Nvidia Says Its Blackwell Chip Is OK, Nothing to See Here

A large part of Nvidia’s growth this quarter was driven by data center revenue, which reached $30.8 billion for the quarter, which was up 112 percent from a year ago. The company’s gross margin was 74.5 percent, actually down from a year ago. But analysts expect Nvidia’s margins could shrink as the company switches to producing more Blackwell chips, which are more expensive to make than their more advanced predecessors.

Nvidia’s earnings reports are seen as important in the AI ​​industry. Chip makers’ advanced GPUs, which enable complex neural network processing, are what made the current productive AI boom possible. As Silicon Valley giants raced to create new chatbots and imaging tools over the past few years, Nvidia’s revenue exploded, allowing it to overtake Apple as the world’s most valuable public company. Since the launch of ChatGPT in November 2022, Nvidia’s stock price has risen almost tenfold.

Almost all major technology companies working on AI, even those that build their own processing units, rely heavily on Nvidia GPUs to train their AI models. Meta, for example, said it builds its latest AI technology on a cluster of more than 100,000 Nvidia H100s. Smaller AI startups, on the other hand, have been left without sufficient AI computing power as Nvidia struggles to keep up with demand.

Blackwell, Nvidia’s new GPU, is made of two pieces of silicon each the same size as its previous chip, Hopper, fused together into a single piece. This design has resulted in a chip that is said to be four times faster and has twice the number of transistors as its predecessor.

But Blackwell’s launch didn’t go well. Originally scheduled to ship in the second quarter, the new chip ran into a production bottleneck, which reportedly delayed the release by several months. Huang addressed this problem, calling it a “design error” that caused “low yields.” Huang told Reuters in August that Nvidia’s longtime chipmaking partner, Taiwan Semiconductor Manufacturing Company Limited, helped Nvidia fix the issue.

Moorhead told WIRED that he remains active at Nvidia and is confident that the generative AI market will continue to grow for at least the next 12 to 18 months, despite some recent reports suggesting that AI progress is starting to pick up.

“I think the only way shareholders can rebel is if they’re worried about the big costs or profits of hyperscalers,” Moorhead said, referring to big tech companies such as Amazon, Google, Microsoft, and Meta in which they are heavily invested. AI cloud services. “But I think they will continue to buy Nvidia until that day comes.” Enterprise AI is still a growth area for Nvidia as well, he added.

On today’s earnings call, Nvidia chief financial officer Colette Kress said that Nvidia’s business AI tools are “full of potential,” including a platform that allows other businesses to build their own AI replicas and agents. Customers include Salesforce, SAP, and ServiceNow, he said.

Huang echoed the same thing later on the call: “We’re starting to see business adoption of agent AI,” he said. “It’s really the latest rage.”