Nvidia’s stock has risen 12-fold since ChatGPT arrived in November 2022. In October it carried the company past $5 trillion, a level no corporation had ever touched, even if only briefly.
CNN Business traced the climb to demand. Global companies cannot get enough of Nvidia’s AI chips, and the International Data Corporation puts the company’s share of data center chip revenue at 81%. Sales and profits in the October quarter rose more than 60% from a year earlier, beating Wall Street’s expectations. Nvidia expects record sales in 2026, projecting around $500 billion in revenue.
From video games to AI training
How it works is a story of a chip finding a second job. Jensen Huang founded Nvidia in 1993 to make graphics processing units for video games. Those GPUs later proved well suited to training AI models. The breakout came only after ChatGPT appeared in late 2022. In 2023, three decades after its founding, Nvidia became the ninth company to reach a $1 trillion valuation. It has shed some value in recent months but remains the world’s most valuable company.
The chips are only part of the pitch. Nvidia also sells full server racks packed with other chip types, plus software that lets developers tune their code to its hardware. It is trying to lay the groundwork for data centers of the future, which it calls “AI factories,” and to move into robots, quantum computers and self-driving cars. “Our vision is that, someday, every single car, every single truck will be autonomous,” Huang said.
Deals worth billions
Much of 2025 went on striking deals. In April, Nvidia said it plans to build $500 billion of AI infrastructure with partners in the US. In September it put $5 billion into Intel to develop custom data centers. Days later it announced a partnership with OpenAI, under which it will invest up to $100 billion in expanding AI data center capacity built around its chips. Reports of friction between the two companies followed, but executives at both have said they remain committed.
Abroad, Nvidia is working with telecoms and governments in France, Germany, Italy and the United Kingdom on sovereign AI infrastructure. In South Korea it is collaborating with government and businesses to deploy over 26,000 chips.
The new Vera Rubin chip, unveiled last month and widely seen as the next growth driver, is in production and due in the second half of 2026. Microsoft, Amazon Web Services, Google Cloud and CoreWeave will be among the first to deploy it.
Pressure points
Rivals such as AMD are pushing in, and some customers are building their own chips to depend less on Nvidia. Fears of an AI bubble persist, a worry Huang has repeatedly brushed aside. In China, US tech controls have restricted exports of its most advanced chips. Huang persuaded President Donald Trump to reverse some limits and allow sales of its second most powerful chips, but Beijing has not confirmed that its companies may buy them.
