Nvidia continued to see demand for its graphics processors in the emerging world of artificial intelligence in its fourth quarter earnings reported Thursday.
In its fourth quarter earnings release, the Santa Clara, Calif.-based company reported revenue of $2.17 billion, up 55% year over year, on earnings per share of $1.13, up 117% a year ago. Wall Street analysts estimated $2.11 billion in revenue on EPS of 83 cents.
Traditionally, the company's processors have been mostly used to power the latest gaming graphics, but the chips have become popular to run AI software in the data center and autonomous vehicles. A specific branch of AI, called deep learning, is where Nvidia's processors particularly shine. These new growth areas have resulted in Nvidia's stock blowing up 360% over the past 12 months.
“Deep learning on Nvidia GPUs [graphics processing units], a breakthrough approach to AI, is helping to tackle challenges such as self-driving cars, early cancer detection and weather prediction," said Nvidia cofounder and CEO Jen-Hsun Huang in a statement. "We can now see that GPU-based deep learning will revolutionize major industries, from consumer internet and transportation to health care and manufacturing. The era of AI is upon us."
But with such heady growth, some investors have begun wondering how much juice is left. The company's data center business appears safe for now.
Data center revenue grew to $296 million, up 205% year over year. And its increasing faster over time: Last quarter, reported in November, Nvidia's data center business showed 193% growth in year-over-year revenue. Google, Amazon and Microsoft are buying more of Nvidia's chip for their cloud services. Nvidia also recently started working with the National Cancer Institute for the United States Department of Energy to develop an AI framework for advanced cancer research.
Analysts warn that growth in the data center may slow in the coming year. Tech giants like Google buy lots of Nvidia chips, but the company is starting to use more of its own custom chip, called the Tensor Processing Unit, in its data center infrastructure. Intel also acquired AI startup Nervana last year to develop a custom AI processor due out later this year.
"I think the general trend for the use of GPUs in the data center is definitely still accelerating in the near term," said Mark Hung, vice president of research at Gartner. "But what we'll see, starting in the second half of this year, is more purpose-built chipsets for machine learning workloads."
Huang is dismissive of some of these efforts and the impact it could have on his business. "TPUs and Nervana are nonstarters," he said in an interview following the earnings call. "The reason for that is they're too application specific. It's like if a CPU [central processing unit] could only do deep learning. GPUs can be used for almost everything."
Huang pointed out that the company's data center business relies on other areas outside of deep learning applications, including high-performance computing and running graphics-heavy software from the cloud.
Automotive growth appears to be cooling down with revenue of $128 million, up 38% year over year, for the fourth quarter. Although growth isn't explosive in auto, most of these autonomous driving efforts are still in the very early pilot stages, so it may take a while for growth to start showing up. Nvidia recently announced autonomous driving partnerships with carmakers like Audi and Mercedes-Benz, and autosuppliers like Bosch and Germany's ZF.
Meanwhile, Nvidia's biggest business, gaming, continued to grow with $1.35 billion in revenue, up 66% year over year.
Despite beating estimates, Nvidia's shares were trading down nearly 2% in after-hours trading. The company's revenue forecast for the current quarter isn't much better than forecasts from analysts: Nvidia is expecting $1.90 billion in revenue, versus $1.88 billion analysts were estimating.
