Nvidia beats expectations with US$39.33B revenue for Q4 FY25, reinforcing AI dominance

US: Nvidia reinforced its dominance in the artificial intelligence (AI) market after reporting US$39.33 billion (S$52.76 billion) in fourth-quarter revenue, surpassing Wall Street’s expectation of US$38.05 billion. The company’s adjusted earnings per share stood at US$0.89, above the forecasted US$0.84. However, its shares were flat in extended trading, according to CNBC.

During the quarter, the company’s net income rose to US$22.09 billion, or US$0.89 per share, up from US$12.29 billion, or $0.49 per share, in the same period last year. 

However, its gross margin for the quarter was down three points from a year ago at 73%. The company attributed the drop to higher costs for its newer, more complex data centre products.

Nvidia expects around US$43 billion in first-quarter revenue for the fiscal year 2026, plus or minus 2%, beating LSEG’s US$41.78 billion estimate. This would mean a 65% annual increase, although lower than the 262% growth seen in the same period last year.

As AI continues to take over, Nvidia’s revenue continues to surge as demand for its data centre graphics processing units (GPUs) remains strong. Its quarterly revenue jumped 78% year-on-year (YoY), while full-year revenue climbed 114% YoY to US$130.5 billion.

While Nvidia’s sales more than tripled in the fourth quarter of fiscal year 2024 and it became the world’s most valuable company, overtaking tech giants Apple and Microsoft in June last year and again in late January this year, CNBC reported that its growth is slowing as it expands.

This year, the focus is on how quickly Nvidia can deliver its next-generation AI processors, known as Blackwell.

The company’s latest AI chip generated US$11 billion in sales during the fourth quarter.

In the company’s press release, Nvidia founder and CEO Jensen Huang described the demand for the chip as “amazing.” He added, “We’ve successfully ramped up the massive-scale production of Blackwell AI supercomputers, achieving billions of dollars in sales in its first quarter. AI is advancing at light speed as agentic AI and physical AI set the stage for the next wave of AI to revolutionize the largest industries.”

Nvidia’s CFO Colette Kress stated that the company expects to see a “significant ramp” in Blackwell sales in the first quarter and called it the “fastest product ramp” in the company’s history. She said in a statement that Blackwell sales were led by large cloud service providers that represented around 50% of the company’s data centre revenue.

Sales of Blackwell and Hopper AI chips were reported under Nvidia’s data centre business, which now makes up 91% of its total revenue, up from 83% a year ago and 60% in 2023. Over the past two years, data centre revenue has grown about tenfold.

In the fourth quarter, Nvidia reported US$35.6 billion in data centre revenue, a 93% YoY increase, surpassing StreetAccount estimates of US$33.65 billion.

Nvidia officials said that while its chips were previously used to train AI, its new chips, like Blackwell, would be used for inference, which delivers AI software.

Despite growing AI competition, with Chinese AI startup DeepSeek using fewer Nvidia chips than its competitors, Nvidia remains confident that demand for its chips will continue to rise.

For AI to think and reason, Ms Kress said, it can require 100 times more compute per task than one-shot inferences. Mr Huang added that next-gen AI algorithms could eventually need millions of times more computing capacity.

When asked whether custom chips developed by competitors Amazon, Microsoft, and Google pose a threat to Nvidia, he said, “Just because the chip is designed doesn’t mean it gets deployed.” /TISG

Read also: Nvidia confident in chip demand despite 17% shares drop amid DeepSeek’s struggles with user surge