Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Record sales of artificial intelligence chips sent Nvidia’s revenue soaring 262 per cent in the past quarter, beating investors’ lofty expectations and pushing shares higher in after-hours trading on Wednesday.
Revenue for the three months to the end of April was $26bn versus consensus estimates of $24.7bn. The enormous year-on-year increase was similar to the prior quarter, when growth hit 265 per cent. For the current quarter it expects revenue of about $28bn, plus or minus 2 per cent, compared with consensus estimates of $26.8bn.
Nvidia shares, which have extended their blistering rise by more than 90 per cent since the start of the year, were up 2 per cent in after-hours trading. The chipmaker also announced a 10-for-one stock split effective from June 7 and said it was raising its quarterly cash dividend 150 per cent.
Ahead of the results announcement, traders were bracing for big swings in Nvidia shares and markets more broadly. The stock’s massive rally has made it one of the most closely watched names on Wall Street. Since the start of 2023 its market capitalisation has increased more than six-fold to $2.3tn, overtaking Google parent Alphabet and Amazon to become the third most valuable US-listed company.
Demand for Nvidia’s AI data centre graphics processing units has skyrocketed in the past year as the biggest tech companies rush to develop the computing infrastructure needed to deliver powerful new AI products at scale. Google, Microsoft, Meta and Amazon have all indicated that their spending would remain high throughout 2024.
Nvidia’s data centre revenue, which relates to its coveted AI chips, was up 427 per cent year on year for the quarter to $22.6bn.
Nvidia has been moving quickly to capitalise on the surge of demand for AI and stay ahead of competitors and customers which are developing their own AI chips. It revealed its new Blackwell chips in March, which it says are twice as powerful as the current generation of chips for training AI models, and offer five times the performance on “inference” — the speed at which such models can respond to queries. That came just a year after the company revealed its previous generation of GPU chip architecture, known as Hopper. Blackwell is expected to start shipping later this year.
Analysts had questioned whether the transition to a new product line could affect the massive year-on-year growth that Nvidia recorded in prior quarters, as a temporary “air pocket” in demand emerges. The rapid pace of its new chip releases has for example led Amazon to change plans to order chips based on the last generation of Nvidia’s architecture and replace them with the new Blackwell line.
Diluted earnings per share were $5.98, up more than 600 per cent from a year ago. Gross margin was 78.4 per cent, slightly better than the 77 per cent analysts had predicted, with net income at $14.9bn, ahead of expectations for $13.2bn.
Rivals AMD and Intel have been rolling out AI data centre chips to compete with Nvidia’s, as well as joining forces with Nvidia’s clients to offer alternatives to its software platform, Cuda, which cements its dominance as a chip supplier.
In April Intel and AMD announced lacklustre first-quarter results and modest guidance, suggesting they are yet to reap the benefits of the explosion in demand. On Tuesday, Microsoft announced it would be using AMD’s new MI300X accelerator chips and its ROCm software to run some of the most demanding AI workloads on its Azure cloud service.
“Nvidia beat on the data centre [revenue] and beat across the board,” said Daniel Newman, chief executive of The Futurum Group. “The whole market was waiting for this number and Nvidia delivered.”
The stock split would create “more accessibility” as well as “additional momentum for the stock”, he added. “The AI trade is alive and well.”