After a quarter where Nvidia’s sales almost doubled, investors and analysts wonder how long the chipmaker can keep this kind of growth now has $ 140 billion per year revenue run rate.
That hope rests on Blackwell, which is Nvidia’s name for a family of server products based on next-generation AI chips.
CEO Jensen Huang and CFO Colette Kress gave investors some new data points on how Blackwell is performing on a call with analysts on Wednesday. The duo confirmed that the rollout is on track, and they hinted that Blackwell’s sales over the next few quarters will be limited by how many chips and systems Nvidia can make, not how much they can sell.
“Blackwell production is at full steam,” Huang said. “We will be shipping this quarter more Blackwells than previously thought.”
The company’s positive comments on Blackwell were one of the reasons why the stock was only down 1%, even though the company did not have higher expectations from bullish investors who expect Nvidia to beat its own forecasts.
Huang and Kress’ comments also addressed fears about shipping delays fueled by reports that Nvidia is making engineering changes to the system to address the issue.
Some of Nvidia’s most important end customers have already received some Blackwell chips, the company confirmed on Wednesday. Microsoft, Oracle and OpenAI have posted images of Blackwell-based server racks on their social media accounts, and on Wednesday, the company said 13,000 Blackwell chips had been shipped to customers.
“There’s still a lot of engineering going on at this point,” Huang said. “But you see from all the systems that are standing, Blackwell is in good shape.”
The sample chips aren’t the bulk of shipments the company wants. This is an early version intended to allow customers to begin testing and preparing their systems and software for volume shipments, which Nvidia plans to begin in the current quarter.
“We will ship more Blackwells next quarter than this (quarter), and we will ship more Blackwells this quarter than from the first quarter,” Huang said.
In July, Nvidia said it expected “several billion dollars” of Blackwell’s revenue in the current quarter, and on Friday, the company said it expects Blackwell’s sales numbers for this quarter to be higher than the original forecast. Huang also said Microsoft will begin previewing Blackwell-based systems to cloud customers.
The limiting factor in producing more Blackwell systems is the number of components Nvidia’s suppliers can provide, Huang said. Additionally, it takes time to increase the speed of manufacturing processes that have gone from zero shipments to billion dollar shipments in a matter of months.
“It’s a case of demand outstripping our supply, and that’s expected when we start this generative AI revolution,” Huang said.
He also named some of Nvidia’s “great partners,” including TSMC, Amphenol, Vertiv, SK Hynix and Micron.
“Almost every company in the world seems to be involved in the supply chain,” Huang said.
Nvidia said that Blackwell’s gross profit will be lower in the coming months than the 73.5% reported in the third quarter, but the company said that the margin will increase as the product matures. Huang indicated that the Blackwell is only a chip by itself or in a configuration that includes all the racks and other components.
Nvidia’s overall message on Wednesday was that its new Blackwell chips are lacking because companies like OpenAI need the fastest GPUs available quickly to develop next-generation AI models. When Blackwell launches, Nvidia’s current AI chip, called Hopper, will be relegated to serving AI models, not creating new ones. Nvidia says sales of the Blackwell will outpace the Hopper.
“You see now that on the back end of the last generation base model, we’re at about 100,000 Hoppers,” Huang said. “The next generation starts at 100,000 Blackwells.”
WATCH: Nvidia can still grow even as Amazon and Microsoft enter the space: Susquehanna’s Chris Rolland
