“Going forward, data centers will dedicate most of capex [capital expenditure] to accelerated computing and AI,” Huang said. “Data centers will increasingly become AI factories, and every company will have them, either rented or self-operated.”

Wednesday marked the company’s first earnings release since DeepSeek debuted a large language model that could match those of American competitors like OpenAI for a fraction of the cost, supposedly using reduced-capacity Nvidia chips. Huang’s company lost nearly $600 billion in market cap on the news, the largest single-day drop for any U.S. firm in history. The chipmaker’s major customers, however, don’t seem deterred, with Meta, Amazon, Google, and Microsoft set to invest as much as $320 billion in AI and data center build-out, per CNBC, based on comments from these companies’ CEOs on earnings calls earlier this year.

Echoing previous comments, Huang said demand for Nvidia’s next-generation Blackwell offering is “extraordinary.” The company said it had sold $11 billion worth of the chips for the quarter.

“Customers are anxious and impatient to get their Blackwell systems,” Huang said.

Blackwell moves the needle by allowing customers to customize data center configurations to fit their needs, but the platform originally faced delays amid reports of server racks overheating. At the time, Nvidia said those issues were “normal and expected,” but Huang acknowledged them on Wednesday’s call.

It proved difficult for the company to switch from focusing on the company’s Hopper chips, which initiated the AI boom, to Blackwell, he told analysts. Future updates, he said, will be easier to implement.

“This is the fastest product ramp in our company’s history, unprecedented in its speed and scale,” CFO Colette Kress said.

Huang says DeepSeek won’t dampen demand

Huang waited until the end of the call to directly acknowledge DeepSeek’s R1 model, which he said had “ignited global enthusiasm.”

“It’s an excellent innovation, but even more importantly, it has open-sourced a world-class reasoning AI model,” he said.

Earlier on the call, he emphasized that demand for compute power and cutting-edge chips is only scratching the surface. He cited the rise of digital assistants, which can use AI to complete complex, multistep tasks, as well as AI’s integration with robotics.

“Each one of these [is] barely off the ground,” he said of both trends, “and we can see them.”

Huang also used similar logic when talking about the effect of U.S. export controls on Nvidia’s sales in China. He acknowledged, however, that revenues in the country had been cut in half since the administration of former President Joe Biden initially put those restrictions in place.

“No technology has ever had the opportunity to address a larger part of the world’s GDP than AI,” he said.

Any obstacle, big or small, needs to be evaluated in that context, he added. For now, it appears tech bulls can breathe a sigh of relief.