Chip giant Nvidia is starting to capitalize on the craze for language-generating artificial intelligence, projecting a more than 64% jump in sales as the company rushes to get more processors in customer hands to satisfy booming interest in the technology.
A new generation of advanced Nvidia chips for AI calculations in data centers is in production, Nvidia CEO Jensen Huang said, and “we are significantly increasing our supply to meet surging demand for them.”
The company forecast a record $11 billion in sales for the current quarter, far above the $7.2 billion Wall Street was expecting and what would be the highest quarterly total ever for the company.
“This demand has extended our data center visibility out a few quarters and we have procured substantially higher supply for the second-half of the year,” Chief Financial Officer Colette Kress said on an earnings call.
Nvidia’s shares, which have more than doubled in value this year, surged more than 28% in after-market trading to reach an all-time high. The rise puts Nvidia, the U.S.’s largest chip-supplier by market value, close to becoming the world’s first $1 trillion chip company.
Demand for computing power that drives language-generating tools such as OpenAI’s ChatGPT is opening a huge new revenue opportunity for the company and others. It has spurred an arms race between tech giants to offer advanced AI features to their customers. Microsoft, which has invested in OpenAI, has been adding the technology to its Bing search engine and business software products. Google has introduced its own advanced AI tools. Facebook parent Meta Platforms also has been working on the technology. Nvidia’s chips are essential in creating these kinds of tools, analysts say, and building just one such AI system can require thousands of Nvidia’s computing engines.
Huang said the company was well-prepared to benefit from the AI opportunity because it was starting to produce a new wave of advanced equipment for data centers when the explosion of interest began last year. “I call it the iPhone moment,” he said, referring to the shift toward smartphones that Apple capitalized on by releasing its advanced handset about 16 years ago. “All the technology came together and helped everybody realize what an amazing product that can be and what capabilities it can have.”
Nvidia doesn’t manufacture its own chips, but farms out production to contract chip-makers including the world’s largest, Taiwan Semiconductor Manufacturing Co. TSMC’s shares, which trade in New York and Taiwan, rose by 7% after-hours in the wake of Nvidia’s results.
Nvidia may be the leading provider of AI chips, but Huang said the battle to supply chips to satisfy demand is fierce. “We have competition from every direction,” he said, from established semiconductor companies to startups.
Nvidia on Wednesday said revenue fell 13% to $7.2 billion in its last fiscal quarter, topping forecasts from analysts surveyed by FactSet. Net profit rose 26% to $2 billion. The sales retreat was driven by a sharp decline in the graphics chips business for videogamers, who pulled back after the pandemic eased and are only beginning to resume buying.
Huang said operators of big data centers are retooling their computing infrastructure to better address the opportunities offered by AI, creating surging demand for its chips.
“A trillion dollars of installed global data center infrastructure will transition from general purpose to accelerated computing as companies race to apply generative AI into every product, service and business process,” he said.
Nvidia’s data center revenue rose to $4.28 billion in its latest quarter, a record, which Kress said reflected strong demand from consumer internet companies and cloud-computing companies.
Nvidia has said it is working on generative AI with Amazon.com, Microsoft and Alphabet’s Google unit, and is partnering with cloud-computing companies to help make generative AI available to smaller businesses. The company on Tuesday said it was adding its AI software to Microsoft’s Azure cloud-computing service, allowing corporate customers to tap in to its chips and software to speed up large generative AI systems.
Nvidia has its roots in graphics-processing chips for videogamers, but has diversified its customer base rapidly in recent years. Engineers found the chips to be well-suited to AI tasks and cryptocurrency mining, which led to an explosion of new sources of demand.
The company has tried to capitalize on that shift by making specialized chips for those markets. Its AI chips have helped its data center division surpass its gaming division in revenues over the past few quarters, a major break from the past. The company has recently begun to roll out a new generation of AI chips for data centers that promise a substantial performance upgrade, and many customers have had to wait for them amid red-hot demand.
The company’s gaming division fell by 38% to $2.2 billion in the latest quarter, which Kress attributed to the macroeconomic slowdown and Nvidia limiting shipments so customers run through existing inventories of chips.
Amid the growth in AI-related sales, new U.S. regulations seeking to hamstring China’s AI industry limited the sale of Nvidia chips there. Nvidia has developed versions of its chips that don’t exceed performance thresholds, but the company said last year that the curbs could cost it up to $400 million in quarterly sales.
