Nvidia partners with Google Cloud to boost AI for 5G app dev

Nvidia has partnered with Google Cloud to create its first AI-on-5G Innovation Lab to help developers more quickly deliver applications for both technologies.

The company also added support for Arm processors in its next generation Aerial A100 AI-on-5G platform. The move is meant to help users deliver a greater number of AI services to the edge through less expensive industry standard servers that combine Arm processors and Nvidia's AI software.

Nvidia believes such servers offer a more straightforward path to building and deploying self-hosted virtual radio access networks (vRAN) by delivering AI and 5G across the span of private enterprises, network equipment companies, software developers and telecommunications services providers.

"With support for Arm, our latest Aerial platform is in a position now to accelerate AI-on-5G," said Ronnie Vasishta, senior vice president of telecom at Nvidia.

The deal is the latest example of hyperscalers including Google, AWS and Microsoft trying to work with and eventually compete with telecommunications providers, said Alan Weckel, founder and technology analyst with the 650 Group, LLC in Incline Village, Nevada.

Microsoft last year acquired both Metaswitch Networks, a company with software that telecommunications companies use to deliver voice and data services, and Affirmed Networks. Amazon struck up a partnership with Verizon in late 2019 to improve 5G speeds.

Google has not been as active in the telco industry as its archrivals until recently, Weckel said.

"Google is now playing in that [telco] space and that's really what the Google labs announcement is about," Weckel said. "Software developers are still trying to figure out whether you develop for the 5G cloud and run it on a Google or an AT&T," he said.

The new lab gives corporate and third-party developers access to Google Cloud's Anthos platform and Nvidia's server hardware and software to improve performance, operational efficiency and reliability. The two companies expect to begin development sometime in the second half of this year.

The Nvidia Aerial A100 AI-on-5G computing platform uses the Nvidia Aerial software development kit and will incorporate 16 Arm Cortex A78 processors into its BlueField-3 A100 data processing unit (DPU). This combined product results in a self-contained, converged card capable of delivering edge AI applications over cloud-native 5G vRAN.

The Bluefield-3 A100 can be deployed with either the x86 or Arm processors. With this choice available to them, networking professionals can deploy 5G base stations and AI applications more easily while taking advantage of their investments in existing infrastructure.

When you think about 5G coming along, the industry is moving away from dedicated proprietary hardware boxes to a more open platform. This is how telco providers want to create and deliver AI-based services in the future.

Nvidia officials said by 2024 they plan to implement many of the capabilities now contained on the board-level Aerial A100 AI-on-5G product, in a system-on-a chip (SOC) offering.

"In the past, companies like Nokia, Ericsson and Samsung built equipment that run the networks handling things like a radio, antenna and cell towers," said Bob O'Donnell, president and chief analyst at Technalysis Research, based in Foster City, Calif. "Now everyone is moving to standard off-the-shelf servers that run a lot of these network functions in software."

The new offering gives Nvidia a competitive advantage, O'Donnell said, but the competitive landscape won't change overnight.

"As companies build out new 5G networks and want to add edge computing to the mix, this can be a convenient solution because now you can do more things inside a single box," O'Donnell said.

The new support for Arm chips offers another confirmation that telecommunication providers are moving in lock step with their hardware and software partners to more open platforms.

"When you think about 5G and 6G coming along, the industry is moving away from dedicated proprietary hardware boxes to a more open platform and that is what Nvidia has announced here," Weckel said. "This is how telco providers want to create and deliver AI-based services in the future."

BlueField-3 A100 will be available in the first half of 2022.

As Editor At Large with TechTarget's news group, Ed Scannell is responsible for writing and reporting breaking news, news analysis and features focused on technology issues and trends affecting corporate IT professionals. He has also worked for 26 years at Infoworld and Computerworld covering enterprise class products and technologies from larger IT companies including IBM and Microsoft, as well as serving as Editor of Redmond for three years, overseeing that magazine's editorial content.