Nokia and Nvidia Launch the First Commercial AI-RAN Platform for Mobile Networks

What happened?

Nokia and Nvidia have announced the launch of the first commercial AI-RAN (Radio Access Network with artificial intelligence) platform. The platform, called Nokia AI-RAN, combines Nokia's network software with Nvidia's GPUs and AI ecosystem, allowing RAN functions to run on AI-accelerated hardware. According to Nokia, this architecture can double network capacity and reduce energy consumption by up to 50%. The announcement, made at Mobile World Congress 2026, marks a milestone in network virtualization, as for the first time such a solution is commercially available and not just a prototype. According to The Next Web, executives from both companies described this launch as "the biggest change in radio in decades." The platform is based on Nvidia's Aerial architecture, an accelerated computing framework for 5G, and Nokia's RAN software, which now runs on Nvidia H100 and B200 GPUs. This enables real-time radio signal processing with AI algorithms, optimizing spectral resource allocation and reducing interference.

Why is this important?

The RAN is the most expensive and complex part of a mobile network, responsible for connecting user devices to the core network. Traditionally, the RAN relies on specialized hardware (ASICs, FPGAs) and proprietary software, limiting flexibility and making upgrades costly. AI-RAN represents a paradigm shift by virtualizing RAN functions and optimizing them through AI algorithms. This enables dynamic resource allocation, interference reduction, and improved spectral efficiency. Historically, network virtualization (NFV) has been applied to the core network, but the RAN has been more resistant due to strict latency and performance requirements. Now, with Nvidia's GPUs, the necessary real-time processing can be achieved. The importance lies in the fact that, for the first time, such a solution is commercially available, opening the door for telecom operators to adopt AI at the heart of their networks. This could accelerate the deployment of advanced 5G (5G-Advanced) and pave the way for 6G, which is expected to natively integrate AI. According to IDC analysts, AI-RAN could reduce the total cost of ownership (TCO) of networks by 30-40% over five years, by eliminating the need for proprietary hardware and enabling faster software upgrades.

Consequences for the market and users

For operators, AI-RAN promises lower operational (OPEX) and capital (CAPEX) costs, by using standard server hardware with GPUs instead of specialized equipment. Additionally, the ability to dynamically scale according to demand can improve service quality during peak hours. A Nokia study estimates that operators could save up to 40% in energy costs and 30% in maintenance expenses. For end users, this translates into a better experience: higher download speeds (up to 2 Gbps in tests), lower latency (below 5 ms), and greater coverage, especially in dense environments like stadiums or urban centers. It could also facilitate the implementation of use cases such as augmented reality, autonomous vehicles, and smart cities, which require low latency and high reliability. From a market perspective, the Nokia-Nvidia alliance challenges competitors like Ericsson and Qualcomm, which are also developing their own AI solutions for networks. Ericsson has presented its "Ericsson AI-RAN" platform based on Intel hardware, while Qualcomm is betting on integrated chip solutions. The race for AI-RAN is intensifying, and other players like Samsung and Huawei are expected to follow suit. According to a Gartner report, the AI-RAN market could reach $15 billion by 2030, with a compound annual growth rate of 45%.

What readers should know

The platform will be available for testing with operators starting in the second half of 2026, with commercial deployments expected in 2027. Nokia states that AI-RAN is compatible with existing 4G and 5G networks and is designed to evolve toward 6G. The first operators to test the platform include T-Mobile US, Vodafone, and NTT Docomo, according to industry sources. It is important to note that, while the promise is ambitious, mass adoption faces challenges: integration with legacy infrastructure, the need for AI-trained personnel, and cybersecurity of virtualized networks. A GSMA report warns that RAN virtualization increases the attack surface, requiring new security measures. However, the backing of two giants like Nokia and Nvidia lends credibility to the project. Additionally, Nvidia has announced a specific version of its CUDA software for networks, called Aerial SDK, which allows developers to create AI applications for RAN. In summary, commercial AI-RAN is a milestone marking the beginning of the transformation of mobile networks driven by AI, with profound implications for the entire telecommunications industry. Operators that adopt this technology could gain significant competitive advantages in terms of costs and service quality, while end users will benefit from faster and more reliable connectivity. The question is whether the industry can overcome the technical and regulatory challenges to realize this promise in the coming years.