Nvidia CEO Jensen Huang attends the “Winning the AI Race” Summit in Washington D.C., U.S., July 23, 2025.
Kent Nishimura | Reuters
It’s been two years since the explosion of generative artificial intelligence started to transform Nvidia’s business. Since then, the chipmaker’s revenue has more than tripled and profits have quadrupled.
Nvidia‘s fiscal second-quarter earnings report, scheduled for Wednesday, will mark the second anniversary of growth, as the company shifted from being known as a maker of gaming chips to its current position at the heart of the technology industry.
Last month, Nvidia became the first company to hit a $4 trillion market cap, and it’s continued to appreciate in value. Since the end of 2022, around the time OpenAI launched ChatGPT and sparked the generative AI boom, Nvidia’s stock price is up twelvefold. It’s up 33% this year, closing on Friday at $177.99.
Growth is still substantial for a company Nvidia’s size, but it has slowed dramatically. After five straight quarters of triple-digit expansion in 2023 and 2024, growth dipped to 69% in the fiscal first quarter this year. Nvidia is expected to report a year-over-year jump of 53% to $45.9 billion in its second-quarter report, according to LSEG’s consensus of analyst estimates.
Data center revenue in the first quarter accounted for 88% of Nvidia’s total sales, the clearest sign of how significant AI has become to its business. The company said that 34% of total sales last year came from three unnamed customers. Analysts say Nvidia’s top end users are major internet companies and cloud providers such as Microsoft, Google, Amazon and Meta.
“The assumptions and performance of Nvidia really dictates what the market is going to start to price into the AI trade, and that whole AI trade has essentially been driving the market this past year,” said Melissa Otto, head of Visible Alpha Research at S&P Global, which aggregates Wall Street research.
Nvidia makes up about 7.5% of the S&P 500.
Tech’s megacap companies, other than Nvidia, reported quarterly results in late July, updating Wall Street on their investment plans. In all, they’re looking to spend roughly $320 billion on AI technology and data center buildouts this year.
OpenAI, which is still private but has a valuation in the hundreds of billions of dollars, says it will team up with SoftBank and Oracle to spend $500 billion over the next four years on the Stargate project, which President Donald Trump announced in January.
Jensen Huang, co-founder and CEO of Nvidia, displays the new Blackwell GPU chip during the Nvidia GPU Technology Conference in San Jose, California, on March 18, 2024.
David Paul Morris/Bloomberg via Getty Images
Analysts say about half of AI capital spending ends up with Nvidia. The company’s reliance on the so-called hyperscalers leaves it vulnerable to changes in the macroeconomic environment and in the artificial intelligence industry, which remains hard to predict.
OpenAI CEO Sam Altman said last week that he believes “investors as a whole are overexcited about AI,” and even said it could be a “bubble.”
But don’t expect a pullback yet. OpenAI CFO Sarah Friar told CNBC on Wednesday that the company “constantly” doesn’t have enough computing power.
As always, Wall Street will be paying close attention to Nvidia’s guidance and other forward-looking commentary from CEO Jensen Huang. For the fiscal third quarter, analysts are expecting revenue growth of 50% to $52.7 billion, according to LSEG. If Nvidia guides higher and tops estimates for the second quarter, analysts say that kind of “beat and raise” could drive AI optimism even higher.
Blackwell ramp
The most important offering from Nvidia is its Blackwell line, which includes individual graphics processing units and entire systems tying together 72 GPUs.
Strong Blackwell numbers would affirm Nvidia’s continuing technological lead and foothold among its key customers, said Ryuta Makino, an analyst at Gamco Investors, which has a stake in the company.
“It solidifies that hyperscaler spending is still very strong with the Blackwell ramp,” Makino said.
Nvidia said in May that its new product line reached $27 billion in sales, accounting for about 70% of data center revenue. That’s a steep increase from $11 billion in the prior quarter.
As more Blackwell chips get installed, experts anticipate that their superior computing power will enable companies like OpenAI and Anthropic to create even more capable AI models. OpenAI’s GPT-5, which was announced earlier this month, was trained on Nvidia’s last-generation Hopper chips, not the newer Blackwell processors.
Nvidia said last year that Blackwell would be limited by supply — how many chips its partners can build and ship — and not by demand.
Blackwell Ultra is expected to start shipping in the second half of 2025. Nvidia recently pushed back on an analyst report from Asia that said Rubin, the chip technology expected to comprise the bulk of GPU sales in 2027, was seeing early production problems.
One visible sign of Nvidia’s rise is Huang’s worldwide fame. He’s regularly name-checked by Trump and during the quarter traveled to meet with business leaders and officials in Taiwan, China, Germany, England and Saudi Arabia.
Huang recently struck a deal with Trump to regain access to the Chinese market. Nvidia will pay 15% of its China chip revenue to the U.S. government in exchange for licenses to export its China-focused AI chip called the H20, Trump said this month. The president added that he’d asked for 20%, but Huang bargained him down.
The H20 is worth a lot to Nvidia. The chip would have contributed about $8 billion in sales in the second quarter, Nvidia said in May, before the U.S. government said it would require a license to ship it to China, effectively shutting off sales.
Nvidia did not include any H20 sales in its guidance for the second quarter, and analysts doubt that it will include any in its forecast for the current period, partially because the Chinese government is pressuring its cloud providers to use homegrown chips from companies such as Huawei.
If H20 is included in guidance, it could boost revenue expectations by about $2 billion to $3 billion, according to analysts at KeyBanc, who recommend buying the stock. But they said they expect Nvidia to completely exclude it, following Advanced Micro Devices’ lead from early August.
“Additionally, given a potential 15% tax on AI exports and pressure from the China government for its AI providers to use domestic AI chips, we expect management to guide conservatively,” the KeyBanc analysts wrote.
Nvidia is working on a new China AI chip based on Blackwell that would also likely need the president’s approval.
“I’m sure he’s pitching the president all the time,” Commerce Secretary Howard Lutnick said about Huang last week on CNBC’s “Squawk on the Street.”
WATCH: CNBC’s interview with Commerce Secretary Howard Lutnick
