(May 21): Nvidia Corp, facing more investor scepticism, used its latest quarterly report to tout progress in diversifying the company, which aims to rely less on the giant data centre operators that have fuelled its runaway growth.
Though spending has continued to surge from large data centre clients — a group known as hyperscalers — Nvidia predicted that a vast array of other businesses and governments would soon become a bigger source of revenue. They are poised to snap up Nvidia’s chips and other computing products to support their own artificial intelligence (AI) ambitions.
Down the road, so-called physical AI will bring a colossal new opportunity in the form of robots and automated vehicles, chief executive officer Jensen Huang said on a conference call with analysts. “We have got it all covered,” he said.
But investors have become harder to impress. Even after the company beat analysts’ estimates with its results and forecast, the shares slipped about 1% in late trading on Wednesday. Shareholders weren’t swayed by an expansion of investor rewards, including a massive increase to the company’s dividend.
Sales in the three months ending in July will be about US$91 billion ($116.44 billion), the company said in its quarterly report. That topped the average estimate of US$87 billion, though analysts’ projections ranged as high as US$96 billion, according to data compiled by Bloomberg.
See also: Manus weighs raising US$1 bil to unwind Meta takeover — Bloomberg
At the same time, the company is facing the first major challenges to its dominance in AI computing, with a variety of chipmakers trying to carve out a piece of the business. And major buyers of Nvidia’s technology are developing their own in-house components.
Nvidia shares had gained 20% this year heading into the report. That increase outpaced the S&P 500 but trailed most major chip peers.
Nvidia is the top seller of AI accelerators, chips used to develop AI models. But it faces growing competition from across Silicon Valley. Advanced Micro Devices Inc has rival processors, and Broadcom Inc. and Alphabet Inc’s Google are attacking the market with their own technology.
See also: Taiwan seeks to detain three in AI chip smuggling crackdown
Nvidia remains in an enviable position, with Wall Street predicting that the company’s revenue will account for more than a third of the entire semiconductor sector’s sales this year.
“The build-out of AI factories — the largest infrastructure expansion in human history — is accelerating at extraordinary speed,” Huang said in a statement.
Data centre spending, which is the main source of Nvidia’s revenue, hasn’t shown signs of letting up. Hyperscalers plan to shell out a combined total of roughly US$725 billion on AI this year. And based on Nvidia’s latest results, revenue from those companies continues to outpace other sources.
That hasn’t just buoyed sales of accelerators. General-purpose CPUs, or central processing units, also are in greater demand. That’s lifted results for Intel Corp and AMD. Chip upstarts are getting a boost as well: Cerebras Systems Inc, which offers a novel product based on large pieces of silicon, had the year’s biggest initial public offering last week.
Santa Clara, California-based Nvidia doesn’t just sell accelerators. It offers a range of chips, as well as networking, software, AI models and even complete computer systems. That helps make its reach and capabilities unassailable, Nvidia management has argued. The company has said it has more orders than it can fill and is investing to add supply to meet that demand.
In the three months ended April 26, Nvidia’s sales gained 85% to US$81.6 billion. Analysts had estimated US$79.2 billion on average. Profit, minus certain items, climbed to US$1.87 a share. That beat a projection of US$1.77.
The adjusted gross margin, the percentage of revenue remaining after deducting costs of production, was 75%.
To stay ahead of Singapore and the region’s corporate and economic trends, click here for Latest Section
Nvidia boosted its quarterly dividend to 25 cents a share from a penny. And the chipmaker announced US$80 billion in stock repurchases.
Nvidia’s all-important data centre unit generated revenue of US$75.2 billion, compared with an estimate of US$73.5 billion. Networking, part of the data centre division, delivered US$14.8 billion in sales, versus an estimate of US$12.7 billion.
As part of its report, Nvidia said it was transitioning to a new framework that would better reflect “its current and future growth drivers”. Its data centre sales figures will now separate hyperscalers from a group it calls ACIE, for AI clouds, industrial and enterprise customers.
The company is on course to record total revenue of more than US$370 billion this year, according to estimates. By that measure, it will be roughly 22 times the size it was in fiscal 2021. Nvidia easily chalks up more sales in a quarter than its next three largest rivals combined.
Huang has just returned from a trip with US President Donald Trump to China, the largest market for semiconductors overall. US export rules have stymied Nvidia’s growth in that country by restricting sales of AI accelerators on national security grounds.
The Trump administration has begun allowing older Nvidia products to be sold to Chinese customers. But Beijing, trying to cultivate local suppliers, has resisted that initiative. That has left Nvidia mostly locked out of a market that it has said could generate US$50 billion a year.
The company said on Wednesday that it is still not getting any data centre revenue from China.
Meanwhile, Nvidia continues to branch out into new areas. It is beginning to sell general-purpose processors and is offering chips tailored to the inference stage of AI. That is the point where models are already trained and beginning to handle real-world inputs.
The company said it expects to get US$20 billion in CPU revenue this year, which would make it the world’s largest supplier.
Uploaded by Tham Yek Lee
