Could Nvidia become a $10 trillion company? These analysts see a path.

By Britney Nguyen

Some on Wall Street are worried about an AI bubble. But analysts at Cantor Fitzgerald say Nvidia is their top stock pick in chips, as the AI trade is 'just getting started.'

Nvidia CEO Jensen Huang said in August that he expects spending on AI infrastructure to reach $3 trillion to $4 trillion by the end of the decade.

The artificial-intelligence industry "is just getting started," according to analysts at Cantor Fitzgerald. That means big things ahead for Nvidia Corp., which this week became the first public U.S. company to reach a $4.5 trillion market capitalization and which could potentially more than double its valuation, the analysts said.

After the chip maker (NVDA) announced a $100 billion investment in AI startup OpenAI last month, Cantor's C.J. Muse and his team said in a Friday note that the move signals Nvidia "is solidifying its position as the de facto AI infrastructure company, quarterbacking the robust AI industry buildout that is in its infancy today."

The partnership fueled more speculation of an AI bubble and drew concerns over "the circularity of vendor financing," the analysts noted, with critics saying the move showed that Nvidia was investing in companies that would then use the money to buy its chips. But the Cantor analysts said that they are not worried this early in the AI cycle, and that Nvidia is their top pick, with a "clear path" to a $10 trillion-plus market cap.

See more: Why Nvidia's $100 billion investment in OpenAI signals a major transformation

The Cantor team named Broadcom Inc. (AVGO), Micron Technology Inc. (MU), Taiwan Semiconductor Manufacturing Co. (TW:2330) and semiconductor-capital-equipment stocks as the other "prime beneficiaries" of the current AI infrastructure buildout.

Hundreds of billions of dollars are going toward AI infrastructure, and that's something that would have been "impossible to comprehend" a year ago, the analysts said. But the industry has since "entered a new paradigm" with AI reasoning and multimodal capabilities that are leading to a boom in demand for tokens, or the units of data that AI models process. Companies can also charge customers for using an AI model by how many tokens are processed and implement token limits to control usage.

The Cantor team thinks that tokens and AI data centers are still being "underappreciated" by both investors and infrastructure providers. They pointed to "previous construction plans completely understating today's insatiable AI demand," resulting in computing shortages and the current race among companies to build as quickly as possible.

"In a world with compute constraints and power limitations, the ability to generate tokens most efficiently is key to maximizing monetization potential for infrastructure builds," the analysts said, and to them, that means building a full-stack data center "that can act as one unit of compute." Nvidia has all of the components for a full-stack solution, the analysts noted, such as its chips, networking equipment and software ecosystem.

Read: This Nvidia move has sparked criticism, but it could help the stock, one analyst says

Meanwhile, the AI data-center expansion "is existential" for cloud-service providers and countries that want control over AI development and deployment, the analysts said. For cloud companies, there's a risk of "becoming obsolete," they said, while countries could face national-security and economic challenges from being behind in AI.

Nvidia Chief Executive Jensen Huang said on the company's second-quarter earnings call in August that he projects AI-infrastructure spending will reach $3 trillion to $4 trillion by the end of the decade, and the Cantor team said that with "what we are seeing on the demand side, this spend will be required to support token/inference demand." In their view, the return on invested capital "has actually been quite awesome" when looking at OpenAI and its rival Anthropic.

The AI industry is only now "scratching the surface" of agentic AI, or AI systems that can handle tasks autonomously, the analysts said, and with "physical AI as the next frontier," they see "massive amounts of tokens" being generated.

-Britney Nguyen

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

10-03-25 1130ET