Nvidia’s profit for the quarter that ended in July more than doubled from a year earlier to $59.69 billion, the company said Wednesday. Revenue also more than doubled, to $96.22 billion. Just three years ago, Nvidia’s quarterly profit was $6.2 billion.

“AI has reached its inflection point,” Jensen Huang, Nvidia’s CEO, said in a statement.

In an indication that the AI boom still has plenty of momentum, Nvidia told investors that its projected revenue in the current quarter would increase 90% from a year ago to $108 billion. The company expects to grow about 70% next year, even though it is “supply-constrained,” Colette Kress, Nvidia’s chief financial officer, said during a call with analysts and investors.

“Although we will work to close the supply-demand gap, we expect supply to remain a bottleneck” through at least January 2028, Kress said.

Nvidia’s results and projections topped expectations. Wall Street analysts had predicted quarterly profit of $50.94 billion on revenue of $91.96 billion, according to FactSet. For the current quarter, Wall Street predicted revenue of $103.77 billion. Nvidia’s share price was up as much as 4% in after-hours trading on Wednesday.

Nvidia’s chips are the centerpiece of giant AI data centers, and demand for those chips has become a barometer of the AI boom. Other technology companies have been buying tens of billions of dollars’ worth of those chips, making Nvidia the most valuable public company in the world, with a market capitalization of about $5 trillion.

More than a decade ago, Huang pushed Nvidia — which at the time made chips mostly for video games — to make software and chips for building AI. Now, the Silicon Valley company controls an estimated 90% of the market for the advanced semiconductors that power AI.

Over this year and next, Amazon, Google, Microsoft and Meta are expected to spend $1.5 trillion building data centers, according to FactSet. Nvidia’s chips fill those data centers.

In the most recent quarter, Nvidia’s revenue from data centers rose 117% to $89 billion — more than 90% of its sales.

Nvidia and Amazon on Wednesday announced a deal to use 2 million more of Nvidia’s chips on Amazon’s cloud computing service. The companies will also build data centers for the U.S. government, which could allow intelligence agencies to run upcoming advanced AI models from companies like OpenAI.

The race to build data centers has driven a shortage in memory chips that is affecting Nvidia itself. The company is “experiencing extreme pricing conditions in memory,” and its margins will narrow as a result, Kress said. Nvidia will also raise prices in the quarter ending next April.

While Nvidia’s revenue from other sources, like the automotive industry, is “still pretty, pretty small,” those sources could eventually lead to “some valuation expansion,” said David Wagner, the head of equity at Aptus Capital Advisors, an investment firm.

“Data centers are going to be the driver for the near term,” he said. “While the others are growing quite substantially, it’s just not their time to shine yet.”

Nvidia is trying to expand to China, which could deliver billions of dollars in sales for the company. President Donald Trump has permitted Nvidia to sell a powerful chip, the H200, to China. But Huang has been less successful in Beijing, where the government has been slow to allow Chinese companies to buy Nvidia’s chips.

In the quarter that ended in July, H200 shipments to customers based in China accounted for less than 1% of Nvidia’s revenue from data centers, Kress said.

Nvidia is also a lead financier of the AI boom. This month, the company agreed to spend as much as $105 billion to back a data center in Pike County, Ohio, that OpenAI plans to lease. Nvidia and six giant asset managers, private equity firms and banks also announced an effort to raise $500 billion in financing for Nvidia’s customers to pay for chips, data centers and computing power.

The complex financing arrangements are leading some investors to wonder “whether to discount the growth rate and the size of the market” for AI chips like Nvidia’s, said Gil Luria, head of technology research at investment bank D.A. Davidson.

The question for investors, Luria said, is: “How big would the market be if Nvidia weren’t funding its customers?”