Nvidia Data Center Business Posts Record Growth, But Falls Short of Forecast

Plagued by a sharp decline in gaming revenues, Nvidia’s data center business also took a hit last quarter, although less severe. Nvidia reported total revenues of $6.7 billion fell 3% year-on-year and 19% sequentially for the quarter ending July 31. The total revenue was far lower than the $8.1 billion Nvidia had forecast in May.

Nvidia’s lower-than-expected revenues for its fiscal year second quarter reported on Wednesday were not a surprise. During its Q1 earnings report, Nvidia had warned of headwinds and lowered its forecast for Q2. Earlier this month, Nvidia issued preliminary results warning of the shortfall. Nevertheless, the period was markedly worse than the lowered forecast had projected.

A sudden decline in demand for GPUs for gaming PCs contributed most to Nvidia’s woes. To correct excess channel inventory, Nvidia took a $1.22 billion charge. Revenues in that segment of $2.04 billion were down 33% year-on-year and 44% from the previous quarter.

While gaming weighed most heavily, Nvidia’s data center business fell short of expectations despite posting a record last quarter. Nvidia’s data center revenues of $3.81 billion grew 61% compared to the same period last year, but only 1% sequentially.

Challenges

Nvidia’s Jensen Huang

“We were challenged this quarter with a fair amount of supply chain challenges,” founder and CEO Jensen Huang explained during the company’s earnings call.

Huang emphasized that Nvidia doesn’t just sell the GPU chip but includes many other chips and system components, including switches, NICs, and cables.

“These systems are complex,” he said. “Kitting all of the components have to come together for us to be able to deliver the final component.”

Nvidia said despite the shortfall, demand for its data center GPUs remains strong, especially among the hyperscale cloud providers.

Nvidia’s Colette Kress

“Revenue from hyperscale customers nearly doubled year on year,” said Nvidia CFO Colette Kress. “Sequentially, sales to North America hyperscale and cloud computing customers increased, but were more than offset by lower sales to China hyperscale customers affected by domestic economic conditions.”

The Next Wave

Huang and Kress also expressed optimism that a new generation of gaming and data center architectures will boost Nvidia’s fortunes.

“The next wave of computing is coming,” Huang said. “With AI and 3-D graphics advances, developers will extend the internet with virtual world overlays that connect to the physical world.”

Last year’s launch of Omniverse, Nvidia’s platform for creating metaverse applications, is gaining momentum, Huang said.

“We created Omniverse to connect the digital and physical world and to be an open platform for creating and operating Metaverse applications,” said Huang. “The immediate applications for Omniverse span product design, manufacturing — and Omniverse is off to a great start.”

During next month’s GTC developers conference, Nvidia plans to roll out the Nvidia H100 data center GPU. The H100 is the first GPU based on Nvidia’s new Hopper Architecture, named after prominent computer scientist Grace Hopper.

Hopper has a new AI modeling engine called Transformer.

“I fully expect the Hopper to be the next springboard for future growth,” Huang said. “This new model, transformers, can’t possibly be overstated. The impact of this model across robotics, computer vision, languages, biology, chemistry [and] drug design is just really quite spectacular.”

Kress added that the flagship Hopper architecture H100 data center GPU is now in production.

“Top computer makers including Dell HPE, Inspur, Lenovo and Supermicro are adopting the new Nvidia Grace CPU super chip to build the next generation of servers,” she said.