Ahead of MWC Barcelona, Nvidia bets AI-native platforms will carry telecom into 6G

Nvidia Corp. early Sunday announced ahead of the MWC Barcelona conference that its joining global telecom leaders in a commitment to construct 6G on open and secure artificial intelligence-native platforms, bringing software-defined networking to the longer term of telecommunications.

Because it rolled out, 5G helped offer enough bandwith and opportunity to blaze a trail for the much-anticipated “metaverse,” but that hype has been quickly supplanted by one other darling: artificial intelligence. The promise of AI — and pondering machines acting autonomously across networks — have greatly increased potential demands for switching and traffic.

The present evolution, 5G Advanced, will turn early 5G deployments into something more capable and programmable and, frankly, more useful at scale. It represents a bridge that operators can tune using software-defined networking, allowing AI- and machine learning-assisted radio and operations to attain higher energy efficiency and improved coverage capability, while supporting latest devices.

The expectation is that this may eventually grow into 6G, which is predicted to launch commercially around 2030, with initial trials starting as early as 2028.

Other than transforming traditional connectivity, 6G wireless networks will speed up advancements in physical AI, allowing thousands and thousands of autonomous machines, sensors, vehicles and robots to interact with the true world.

“AI is redefining computing and driving the most important infrastructure buildout in human history — and telecommunications is next,” said Jensen Huang, founder and chief executive of Nvidia. “Along with a world coalition of industry leaders, Nvidia is constructing AI-RAN to rework the world’s telecom networks into AI infrastructure in every single place.”

AI-RAN, or artificial intelligence radio access networking, represents the following evolution of telecommunications architecture: incorporating AI-driven networking that may repeatedly evolve through software, enabling real-time intelligence and faster advancement.

Nvidia joined with leading operators and infrastructure providers including Booz Allen Hamilton Inc., BT Group plc, Cisco Systems Inc., Deutsche Telekom AG, Nokia Corp. and T-Mobile US Inc. to construct open and trusted wireless platforms.

“As 6G becomes the backbone of the AI era, telecom will serve because the nervous system of the digital economy, enabling autonomous systems and intelligent industries at scale and unlocking latest value for patrons and businesses alike,” said Srini Gopalan, chief executive of T-Mobile.

Commercializing AI-RAN

To support this onrushing technological wake, Nvidia announced latest AI-RAN collaborations with industry pioneers including T-Mobile US, SoftBank and Indosat Ooredoo Hutchison.

These partnerships bring AI-RAN out of the lab and into the sector, providing platforms for commercialization of products. This features a growing ecosystem around Nvidia-powered solutions reminiscent of Quanta Cloud Technology off-the-shelf systems; WNC Corp.’s AI-optimized indoor-outdoor radio unit; Eridan Communications Inc.’s 4T4R O-RU; and Lite-On Technology Corp.’s accomplished integration for sub-6 GHz units and millimeter wave.

This is able to allow firms to deploy wireless networking for high-capacity, short-range coverage and dense infrastructure in urban environments, as needed, depending on deployment requirements.

Advancing AI network software for telcos

Autonomous networks — intelligent, self-managing telecommunications operations — would mean telecoms operating themselves like pondering machines. That can require large language models and reasoning systems that may automate networks from inside.

Nvidia argues that for networks to grow to be autonomous, agentic AI might want to run using specialized telecom network models that may refer to one another across networks and use simulation tools to validate actions.

In its announcement, the corporate unveiled a Nemotron-based large telco model, or LTM, together with a guide for constructing agents for network operations and an operations blueprint. These Nvidia Blueprints include energy savings, network configuration with multi-agent orchestration and advanced autonomy.

The corporate said its Nemotron framework is open and offers telcos transparency into the way it was trained and what data was used. That, Nvidia argues, enables secure and fast on-premises deployment inside networks.

Nvidia and Tech Mahindra Ltd. published an open-source guide showing operators how you can fine-tune domain-specific reasoning models and construct agents for network operations center workflows.

Photo: ArtisticOperations/Pixabay