|

Nvidia May Take Humans out of the Data Center Equation

The filing signals that the company wants to be a thought leader in the data center space.

Photo of a Nvidia patent
Photo via U.S. Patent and Trademark Office

Sign up to uncover the latest in emerging technology.

Nvidia is putting robots to work. 

The company filed a patent application for “intelligent components of a data center,” which details a system to use autonomous machines to manage and operate data center server racks and components. 

“As the size of data centers and the complexity of their layouts increases, inspecting and repairing computing equipment in this manner becomes more difficult for technicians,” Nvidia said in the filing. 

Nvidia’s tech relies on robots to manage the movement and positioning of server racks. A central monitoring system oversees operations, issuing commands to the robots to reconfigure the racks based on whatever the data center needs at any given moment to function efficiently. The central monitoring system also watches and adjusts the data centers’ cooling systems in real time.

The arrangement could make it easier to scale and adjust data centers based on factors like environmental conditions, computational demand or maintenance needs, without having to involve human technicians.

However, Nvidia noted, if there are problems that robots can’t solve, a technician is “summoned via an alert, a message, or an alarm, to manually investigate the reason for failure.” 

The patent application adds to a broader IP push from Nvidia, seemingly seeking to cement its place as a thought leader in data center architecture, said Trevor Morgan, senior vice president of operations at OpenDrives

While many of its filings tackle the energy efficiency issues that server farms pose as AI development continues at a rapid clip, this one combines the company’s AI and automation expertise with its data center strength, creating a “beautiful merging of robotics and intelligence for the purpose of efficiency,” said Morgan. “It’s optimization across all of the dimensions of a data center.” 

Despite pursuing such patents, it’s unlikely that Nvidia will shift its focus to building and operating its own data centers, Morgan noted, especially given how well its current strategy has been working out for the company: The chip giant once again beat earnings expectations on Wednesday, reporting revenue of $35.08 billion, up 94% year over year. Data center revenue alone for the quarter was $30.8 billion, up 112% from the previous year’s quarter. 

Proving that it’s a thought leader in the data center space does more than generate good press, said Morgan. It could also lead to lucrative licensing agreements and partnerships, as well as allow the company to shape what the future of the data center looks like. 

“[Nvidia] owns so much IP that it doesn’t realistically want to build [data centers] itself,” said Morgan. But in owning these ideas and allowing other firms to use them, “they’re dictating the larger environment in a way.” 

However, the data-center robotics patent in particular begs the question that many enterprises are grappling with as AI-powered automation continues to become the new normal: Is it always better to take the human out of the loop? And where should the line be drawn? 

“If a robot can push the server rack better than you can, what do humans do?” said Morgan. “And is it really cheaper to do that? I do think that those kinds of questions are also behind what Nvidia is doing.”