Nvidia CEO: ‘The Next Major Wave Of AI’ Is In The Enterprise

Jensen Huang says the chipmaker’s channel partners are integral to Nvidia’s new sales push with enterprise customers, which includes the new Nvidia AI Enterprise software suite that enables GPU-accelerated computing on VMware’s vSphere platform. He made the remarks as the company reported new records for quarterly revenue as well as data center and gaming sales for its first quarter.

Nvidia CEO Jensen Huang said the company’s new software and hardware solutions for enterprise customers will drive the “next major wave of AI” as the chipmaker reported new records for quarterly revenue as well as data center and gaming sales for its first quarter.

“This is a really large endeavor and the early indications of it are really exciting, and the reason for that is because our data center business is more than 50 percent vertical industry enterprises already. And by creating this easy-to-adopt and easy-to-integrate stack, it’s going to allow them to move a lot faster,” he said Wednesday during Nvidia’s earnings call for the first quarter of its fiscal year 2022, ended May 2.

[Related: Nvidia’s Manuvir Das: We ‘Mimicked’ VMware For Enterprise AI]

Huang said Nvidia’s channel partners, including value-added resellers and system integrators, are integral to selling the company’s new solutions, which include the Nvidia AI Enterprise software suite as well as the EGX server platform that runs the software. This stands in contrast to Nvidia’s cloud service provider business, which is based on direct sales, he added.

“We sell components directly to them. We support them directly. But there are 10 of those customers in the world,” Huang said. “For enterprises, there are thousands, industries far and wide. And so we now have a great stack and a great software stack that allows us to take it to the world’s market so that everybody could buy more and save more.”

“Buy more and save more” is Huang’s mantra for the total cost of ownership benefits that Nvidia’s GPUs can bring to data centers, and with the company’s new Nvidia AI Enterprise software, which allows enterprises to run GPU-accelerated applications on VMware vSphere, Huang said “there’s no question” that the TCO benefits are “dramatic.”

“Because we sell a component, the TCO of the entire system and all the servers and all the people and the infrastructure and the energy cost has been reduced by ‘X’ factors, sometimes 10 times, sometimes 15 times, sometimes five times,” he said.

Huang said Nvidia’s enterprise software efforts have come a long way from when it used to be just about virtualizing the GPU inside a VMware or Red Hat environment, which has made it possible for multiple users to use one GPU.

With Nvidia AI Enterprise and the company’s other new software products, like Nvidia Omniverse and Nvidia Fleet Comment, the company now has an “end-to-end suite of software that is consistent with today’s enterprise service agreements,” he said.

“It’s consistent with today’s enterprise business models and allows us to support customers directly and provide them with the necessary service promises that they expect, because they’re trying to build a mission-critical application on top,” Huang added.

This means new opportunities for solution providers in the channel, according to the CEO.

“By productizing our software, we provide the ability for our large network of partners — OEM partners, value-added resellers, system integrators, solution providers — [to have] a product that they can take to market,” he said.

Fast Growth For GPUs For Inference, Cryptocurrency Mining And Gaming

Huang made the remarks shortly after Nvidia reported revenue of $5.6 billion for the first quarter of its fiscal 2022, an 84 percent year-over-year increase and $250 million above Wall Street’s expectations. The company’s earnings per share for the quarter were $3.66, surpassing Wall Street estimates by 38 cents.

The Santa Clara, Calif.-based company expects revenue in the second quarter to reach roughly $6.3 billion, which would amount to more than double Nvidia’s revenue from the same period last year. The company said the sequential growth will largely be driven by sales of its gaming and data center products.

Nvidia’s stock price was slightly down by around 0.93 percent in after-hours trading Wednesday.

Nvidia said its data center revenue in the first quarter grew 79 percent year-over-year to $2 billion, and the growth was largely driven by the company’s Mellanox networking products — reversing a previous decline for the area — and the ramp of its Ampere GPU products with both hyperscale customers and customers in vertical industries. Unlike the previous quarter, however, hyperscalers drove more growth than customers in vertical industries.

“Hyperscale customers led our growth this quarter as they built infrastructure to commercialize AI in their services,” Collette Kress, Nvidia’s CFO, said on the earnings call. “Cloud providers have adopted the A100 to support growing demand for enterprises, startups and research organizations.”

Kress added that the company saw record shipments of data center GPUs for AI inference applications, which came not just from sales of the company’s T4 GPU but also its new Ampere-based GPUs — the A100, A10 and A30 — all of which are optimized for both training and inference.

The new record shows that Nvidia’s customers are shifting from CPUs to GPUs for inference, Kress said, which is happening because GPUs can “better keep up with the exponential growth and the size and the complexity of deep neural networks,” plus they have lower latency.

The company’s gaming revenue in the first quarter grew substantially, increasing 106 percent year-over-year to $2.7 billion. The company said desktop and laptop GPU sales as well as system-on-chip products for game consoles like the Nintendo Switch all set new sale records.

Kress said the company believed some of the gaming growth came from cryptocurrency mining, but it was unable to discern to what extent.

However, the company has introduced purpose-built cryptocurrency mining processors, also known as CMPs, while limiting the cryptocurrency-mining capabilities of GeForce GPUs meant for gamers, according to Kress. This means cryptocurrency and gaming customers ideally won’t have to compete for supply, which she said will translate into continued growth for both product segments.

“We are making a significant effort to serve miners with the CMP and provide more GeForce cards to gamers,” she said.

This has already resulted in traction for the new product area: CMP sales accounted for $155 million of the $327 million in OEM and other revenue, Kress said, and they were the primary driver for the segment’s 137 percent in year-over-year growth.

Nvidia’s professional visualization revenue in the first quarter grew 21 percent year-over-year to $372 million, driven by record sales of workstation GPUs for heavy-duty laptops. First-quarter automotive revenue was $154 million, a 1 percent year-over-year decrease.