T-Mobile and Nvidia Complete First AI-RAN Call on Live Network

T-Mobile US and Nvidia completed what Nvidia has described as the first artificial intelligence radio access network call conducted on a live commercial network, a milestone unveiled at Nvidia's GTC Berlin conference as part of an ongoing collaboration between the two companies.

What Happened

T-Mobile US executed a live AI-RAN call on its commercial network infrastructure, with Nvidia present at GTC Berlin to announce the achievement. AI-RAN, or AI-native Radio Access Network, refers to the integration of artificial intelligence directly into the software and hardware layers that manage wireless spectrum and signal processing in mobile networks, replacing or augmenting traditional rule-based systems. Nvidia described the call as a first for any live commercial network globally.

The announcement was made at GTC Berlin, Nvidia's European edition of its GPU Technology Conference, where the company has used the venue to highlight enterprise and telecommunications applications of its hardware and software platforms.

Background

The telecommunications industry has been moving toward what carriers and equipment vendors call Open RAN, an architecture that disaggregates the software and hardware components of traditional base stations, allowing third-party vendors to supply individual elements. AI-RAN takes that a step further by embedding machine learning models into network operations to manage tasks such as interference mitigation, spectrum allocation, and energy efficiency in real time.

Nvidia has been positioning its GPU platforms as central infrastructure for this shift. The company announced an AI-RAN Alliance in early 2024 alongside carriers and equipment makers, with T-Mobile listed among participants. T-Mobile has separately committed to network modernization investments tied to its post-merger integration with Sprint, which closed in 2020 following regulatory approval.

T-Mobile is currently the second-largest wireless carrier in the United States by subscriber count. The company has previously highlighted its mid-band 5G spectrum holdings, acquired through the Sprint merger, as a foundation for advanced network capabilities.

What the Technology Involves

In a conventional radio access network, base station software follows deterministic, rules-based logic to manage how signals are transmitted and received. An AI-native RAN replaces portions of that logic with neural network models that can adapt in real time to changing network conditions, user density, and interference patterns.

Nvidia's contribution to AI-RAN deployments involves its aerial software development kit, which runs on its GPU hardware and is designed to handle the computational workloads involved in AI inference at the radio layer. Running such workloads on a live commercial network, rather than a test or lab environment, represents a distinct operational step, as live networks carry actual subscriber traffic under variable and unpredictable conditions.

Industry Context

Several major carriers and equipment vendors have announced AI-RAN research programs or pilot projects over the past two years, including Ericsson, Nokia, Samsung, and SoftBank. SoftBank and Nvidia have previously demonstrated AI-RAN capabilities in Japan. The T-Mobile announcement marks the first time Nvidia has publicly characterized a deployment as occurring on a live commercial network in the United States.

Quantum computing researchers and telecommunications analysts have noted that AI-RAN adoption at scale will require significant capital investment in GPU-capable base station infrastructure, as well as software development to train and maintain the underlying models. No timeline for broad commercial rollout of AI-RAN across T-Mobile's network has been publicly disclosed.

What Comes Next

T-Mobile and Nvidia have not announced a specific deployment schedule for expanding AI-RAN capabilities beyond the demonstrated call, and further technical details are expected to be disclosed through Nvidia's GTC Berlin conference sessions currently underway.