Most of Nvidia’s growth this quarter came from data center revenue, which totaled $30.8 billion in the quarter, an increase of 112% year over year. The company’s gross margin was 74.5%, about the same as a year ago. But analysts expect Nvidia’s profit margins could shrink as it moves to ramp up production of Blackwell chips, which are more expensive to manufacture than their predecessors.
Nvidia’s earnings report is seen as an important bellwether for the AI industry. Chip architects’ advanced GPUs that power complex neural network processing have made the current generative AI boom possible. Over the past few years, as Silicon Valley giants raced to develop new chatbots and image-generating tools, Nvidia’s revenue has exploded, surpassing Apple to become the world’s most valuable publicly traded company. Since ChatGPT was announced in November 2022, Nvidia stock has increased nearly 10x.
Almost all major technology companies working on AI rely heavily on Nvidia GPUs to train their AI models, including those that build their own processing units. For example, Meta says it is building its latest AI technology on a cluster of over 100,000 Nvidia H100s. Meanwhile, smaller AI startups lack sufficient AI computing power as Nvidia struggles to keep up with demand.
Nvidia’s latest GPU, Blackwell, consists of two pieces of silicon, each the size of its previous chip, Hopper, which are combined into a single component. This design resulted in a chip that was four times faster and had more than twice as many transistors than previous chips.
But Blackwell’s launch wasn’t all smooth sailing. The new chip was originally scheduled to ship in the second quarter, but production stalled and the rollout was reportedly delayed by several months. Hwang accepted responsibility for the problem, claiming it was a “design flaw” that “caused the yield decline.” Huang told Reuters in August that NVIDIA’s long-time chip manufacturing partner, Taiwan Semiconductor Manufacturing Company Limited, helped NVIDIA fix the problem.
Moorhead told WIRED that he remains bullish on Nvidia, despite recent reports suggesting that AI advances are starting to plateau, and that the generative AI market is expected to grow at least 12 to 18 years from now. He said he was confident that the moon would continue to grow.
“I think the only way shareholders will revolt is if they’re concerned about capital expenditures and profitability of hyperscalers,” Moorhead said, adding large sums of money to big tech companies like Amazon, Google, Microsoft and Meta. Referring to his investment, he said: AI cloud service. “But I think they’ll just keep buying up Nvidia until that day actually comes.” Enterprise AI remains a growth area for Nvidia, too, he added.
On today’s earnings call, Nvidia Chief Financial Officer Colette Kress said Nvidia’s enterprise AI tools are at “full throttle,” including an operating platform that allows other companies to build their own co-pilots and AI agents. I said that there is. Customers include Salesforce, SAP, ServiceNow, and more.
Huang echoed the same point later in the call: “We’re starting to see enterprise adoption of agent AI,” he said. “It’s really the latest fad.”
