Tech and Telecom Giants Form Open AI-RAN Alliance
A coalition of tech and telecom giants including Ericsson, NVIDIA, AMD, AT&T, and Verizon has launched the OCUDU Ecosystem Foundation. Hosted by the Linux Foundation, the group aims to accelerate the development of open-source software for AI-driven Radio Access Networks (AI-RAN), with Ericsson joining as a founding premier member to guide the architecture for 5G and early 6G systems.
The drive towards an open AI-RAN model stems from the high costs and lack of flexibility in current RAN deployments, which often rely on tightly integrated, proprietary hardware and software from a single vendor. Disaggregating these components is seen as essential for mobile network operators to innovate and reduce operational expenses. The OCUDU software project was initially funded by the U.S. Department of Defense's FutureG Office and the National Spectrum Consortium. They awarded a contract to AI-native wireless company DeepSig and open-source RAN leader Software Radio Systems (SRS) to build the initial carrier-grade software stack. At its core, OCUDU stands for Open Centralized Unit/Distributed Unit, aiming to create the "Linux of RAN." It provides a foundational open-source code base for the two key computing components of the network, a gap that has previously hindered the commercial deployment of fully open-source RAN systems. This initiative directly complements the broader AI-RAN Alliance, which now includes over 75 members. The Alliance is structured into three working groups: "AI-for-RAN" to improve network efficiency, "AI-and-RAN" to merge AI and network workloads on shared infrastructure, and "AI-on-RAN" to enable new edge services. NVIDIA's involvement is central to this vision, contributing its AI Aerial software libraries and advocating for GPU-accelerated infrastructure that can handle both 5G signal processing and AI applications simultaneously. This approach seeks to transform base stations from simple network cost centers into multi-purpose AI hubs. Ultimately, these collaborative efforts are paving the way for 6G networks, which are being designed to be "AI-native" from their inception. Unlike 5G, where AI is largely an add-on, 6G architecture will have AI deeply embedded to autonomously manage network slicing, predict traffic, and optimize resources in real-time.
