Nvidia eyes 6G with AI-native telco stack

• Nvidia is already trying to insert itself into mobile network infrastructure with an AI-RAN proposition
• Now the AI tech giant is expanding that pitch with a slew of partners and positioning itself as the bedrock of 6G network architecture blueprints
• T-Mobile US and Cisco are among its AI-native network stack collaborators
• At the heart of the pitch, not surprisingly, is Nvidia’s AI Aerial platform for wireless networks  

In its latest effort to make itself an indispensable vendor and partner to the telecom network operator community, AI technology giant Nvidia has teamed up with T-Mobile US, Cisco and others to develop an AI-native wireless network hardware, software and architecture for 6G.

Nvidia’s partners in this particular effort are T-Mobile US, Cisco, independent US-based R&D outfit Mitre, Northern Virginia, US-based Open RAN software developer ODC (ORAN Development Company) and US technology/management consulting firm Booz Allen Hamilton. The companies will focus on “the research and development of AI-native wireless network hardware, software and architecture for 6G”, according to a press release issued to coincide with the opening keynote by Nvidia CEO Jensen Huang at the company’s annual GTC event in San Jose. The initial slew of GTC announcements – almost 20 press releases and nearly as many blogs – covered a broad range of product announcements, partnerships and AI-related developments: You can find them all here.

But we’re going to focus on the telecom angle.  

Nvidia is already pushing hard to insert itself into 5G networks via the multi-faceted proposition being pitched by the AI-RAN Alliance: That concept has some big-name telco supporters, such as T-Mobile US and SoftBank – both co-founders of the AI-RAN Alliance along with Nvidia and others – as well as KDDI and Indosat Ooredoo Hutchison (IOH). 

What AI-RAN doesn’t have currently, but which the Alliance is working hard to attract, is widespread mobile network operator commitment: The alliance has more than 75 members but fewer than 10 are network operators. The idea that AI can be used, provisioned and run at the edge of telco networks is not unattractive to the telco community, but the chief concerns are the potential cost of adding GPU-based systems to networks (especially at a time when capital expenditure – capex – is tight) and the impact on power consumption. Nvidia told TelecomTV during the recent MWC25 event in Barcelona that such concerns are becoming a thing of the past and will be dispelled by imminent developments – see Nvidia preps fresh AI-RAN pitch.

Those developments are clearly yet to come but, in the meantime, Nvidia is looking further ahead and trying to position itself as a key player in the 6G era. In terms of network architecture investments and deployments, this is set to begin towards the end of this decade and most meaningfully after 2030.  

Nvidia’s pitch is that “next-generation wireless networks must be fundamentally integrated with AI to seamlessly connect hundreds of billions of phones, sensors, cameras, robots and autonomous vehicles,” and it’s hard to argue against that, given network operators the world over are currently figuring out the optimum ways to integrate AI into their current networks and operations. 

What Nvidia wants for the 6G era, though, is for AI to be “woven in from the start” and for its AI Aerial platform to be the foundation upon which 6G networks, and the stack that it’s developing with its partners, are built. When Nvidia unveiled AI Aerial in September 2024, it described it as “a suite of accelerated computing software and hardware for designing, simulating, training and deploying AI radio access network technology (AI-RAN) for wireless networks in the AI era” that “offers high-performance and energy-efficient software-defined RAN, improved network experience and new revenue opportunities with edge AI applications to host internal and third-party generative AI applications.” The company has now added new features and functionalities to AI Aerial, including digital twin capabilities, an open-source library of modules for its Sionna neural radio framework, and a test bed – see this announcement for further details about the enhancements and the broad range of companies and organisations that are using AI Aerial for application development. 

According to Nvidia, an AI Aerial-based network will provide enhanced spectral efficiency, operational simplicity, lower operational and improved energy efficiency from an open, secure, unified IT system that can run network and AI workloads in harmony. It’s a compelling pitch that ticks a lot of boxes but will telcos buy it? And by that I mean – will network operators believe the marketing pitch and then fork out the greenbacks? That looks like it’s going to be a challenge for Nvidia with the mobile network operator community in general.

One operator that’s on board, as already mentioned, is T-Mobile US, which is expanding its AI-RAN Innovation Center collaboration with Nvidia, initially announced in September 2024, to include “additional research-based concepts for AI-native 6G network capabilities”, working alongside Nvidia’s new set of telecom sector friends. The telco’s CEO, Mike Sievert, stated that “working with these additional industry leaders on research to natively integrate AI into the network as we begin the journey to 6G will enable the network performance, efficiency and scale to power the next generation of experiences that customers and businesses expect.”

Cisco is taking a “lead role” in the collaboration by providing mobile core, security and networking technologies and expertise (as well as decades of telecom network technology experience), while ODC, which appears to be focusing its efforts on what it calls Open RAN 2.0, will provide “Layer 2 and Layer 3 software for distributed and centralised units of virtual RAN as part of the AI-native radio access stack”. The chairman of ODC’s advisory board is Shaygan Kheradpir, whose previous roles in the industry include CEO at Juniper Networks, CEO at Coriant and CIO at Verizon (in its early days). He is currently the telecom and tech lead at Cerberus Capital, which is ODC’s financial backer.

Booz Allen, meanwhile, will “develop AI-RAN algorithms and secure the AI-native 6G wireless platform,” noted Nvidia. “Its NextG lab will conduct functional, performance integration and security testing to ensure the resiliency and security of the platform against the most sophisticated adversaries. The company will lead field trials for advanced use cases, such as autonomy and robotics.” 

Horacio Rozanski, chairman and CEO of Booz Allen, boasted that the company “has the technologies to make AI-native 6G networks a reality and revolutionise secure communications for an entirely new generation of intelligent platforms and applications.”

Will telcos warm to this new angle on AI-RAN? The proof, as my grandmother used to say, will be in the pudding – which, in this case, will be mobile network operator engagement. And to get that, Nvidia is going to need some very compelling financial and sustainability data and assurances.

- Ray Le Maistre, Editorial Director, TelecomTV