AI-Native Networks Move from Vision to Reality at MWC 2026

At Mobile World Congress (MWC) 2026 in Barcelona, the long-discussed vision of AI-native networks shifted decisively from theory to practice. Instead of reiterating future potential, telecom operators, vendors, and hardware makers delivered results in the form of live field trials, new commercial product launches, and commitments to build next-generation networks rooted in artificial intelligence.

Nvidia and Global Operators Commit to AI-Native 6G Infrastructure
One of the event’s most consequential announcements came from Nvidia, which secured formal partnerships with more than a dozen global technology leaders. Companies including BT Group, Deutsche Telekom, Ericsson, Nokia, SK Telecom, SoftBank, T-Mobile, Cisco, and Booz Allen pledged to develop 6G connectivity using open, secure, and AI-native software platforms. This initiative—supported by cross-national collaborations in the US, UK, Europe, Japan, and Korea—underscores an industry-wide commitment to more intelligent, resilient, and trustworthy network infrastructures.

Nvidia, a founding member of the AI-RAN Alliance (now over 130 companies strong), also announced it had joined the US FutureG Office’s OCUDU Initiative to accelerate open, AI-native 6G architecture. The company launched a set of open-source tools for network operators, including a 30-billion-parameter Nemotron Large Telco Model (LTM) co-developed with AdaptKey AI, as well as technical blueprints for AI-based network energy efficiency and configuration.

Nokia and Partners Demonstrate Live AI-RAN Deployments
Nokia, in partnership with Nvidia, reported significant progress, including live, over-the-air functional tests of its anyRAN software running on Nvidia’s GPU-powered platforms. The demonstrations involved T-Mobile US, Indonesia’s IOH, and SoftBank and showcased concurrent use of AI and radio access workloads on the same infrastructure—highlighting the maturity of AI-powered networks beyond the laboratory. SoftBank showcased how spare network compute resources could be harnessed for new, revenue-generating AI services.

Nokia also expanded its hardware partnerships, now involving Dell Technologies, Quanta, Supermicro, and Red Hat OpenShift. This broadens choices for telecom operators deploying AI-based networks using standard, commercial hardware.

Ericsson Advances with Custom AI-Ready Radio Hardware
In contrast to the GPU-centric approach adopted by Nokia and its collaborators, Ericsson introduced ten new AI-ready radios built on its own custom silicon, which feature embedded neural network accelerators. Ericsson’s solution delivers improved energy efficiency and supply chain independence by avoiding reliance on external GPUs. The company also deepened its partnership with Intel, targeting compute and cloud technologies for AI-native 6G. This reflects a key industry debate: whether AI inference should be embedded with custom silicon or offloaded to general-purpose GPUs.

Operators Outline Strategic Rebuilds Around AI
Operators like SK Telecom and SoftBank provided further evidence of AI’s central role in future networks. SK Telecom announced plans for an end-to-end AI-native infrastructure upgrade, including a sovereign AI foundation model with over a trillion parameters and a new AI data center in Korea developed in cooperation with OpenAI. SoftBank, collaborating with Northeastern University and other partners, demonstrated intent-driven autonomous AI-RAN systems capable of automatically translating business goals into real-time network configurations.

Hardware Ecosystem for AI-RAN Expands
Hardware vendors such as Quanta Cloud Technology, Supermicro, MSI, Lanner Electronics, and AMD presented products specifically designed for AI-RAN use cases. The resulting ecosystem now allows operators to run AI inference and core network functions at the network’s edge, giving enterprises new flexibility for managing data and workloads closer to users.

Wider Industry Impact
The shift to AI-native networking has deep implications for how telecoms deliver and monetize connectivity. Continuous, software-driven evolution could make network infrastructure as flexible as cloud computing, while new architectures offer avenues for running enterprise AI workloads on edge hardware. The ongoing debate between GPU-centric and custom silicon approaches will shape future procurement and vendor strategies.

MWC 2026 demonstrated that AI-native networks are no longer a research project but an emerging commercial reality—transforming the industry from the ground up.

Source: artificialintelligence-news.com