KDDI Corporation and KDDI Research have introduced a collaborative initiative focused on building a high-fidelity RAN Digital Twin, a virtual representation of operational radio access networks designed for the 6G era. NVIDIA, Keysight Technologies, and Samsung Research America have joined the project to create an environment where artificial intelligence can be trained, tested, and evaluated safely while supporting future autonomous network operations. The shared platform is intended to accelerate AI-based network optimization by recreating real-world network behavior inside a virtual environment instead of relying on active commercial infrastructure.

The participating companies plan to combine advanced simulation technology capable of reproducing complex wireless behavior with scalable computing infrastructure. This approach allows extensive AI experimentation across numerous simulated scenarios while avoiding risks associated with testing directly on operational networks. The initiative also creates an environment where future network conditions, traffic variations, and radio propagation changes can be evaluated before deployment. KDDI and KDDI Research also expect the platform to support continuous AI algorithm development as network demands increase during the transition toward 6G.

Before launching this collaboration, KDDI and KDDI Research had already developed technology allowing multiple AI agents to cooperate in optimizing base station coverage. That technology is currently being introduced across commercial networks throughout Japan as part of broader efforts to achieve AI-driven autonomous network operations. As expectations for communication quality continue to grow in future wireless generations, the companies identified the need for more advanced AI training methods capable of handling increasingly complex operational requirements without compromising existing services.

The collaboration outlines two principal applications for the RAN Digital Twin. The first focuses on providing a dedicated training and evaluation environment for AI-powered network optimization, allowing developers to simulate thousands of operating conditions while examining potential future scenarios before implementation. The second application supports laboratory-based testing of new AI-enabled RAN capabilities, including technologies associated with the AI air interface. This testing environment enables developers to verify algorithms across diverse virtual conditions before field deployment, supporting quality assurance while reducing development timelines and enabling faster delivery of new AI capabilities.

The project roadmap calls for a scalable prototype by the end of March 2028, followed by broader expansion through the end of fiscal year 2030. During that period, the companies plan to validate performance using KDDI's commercial network while extending the platform to additional use cases. KDDI and KDDI Research also intend to operate the RAN Digital Twin using AI factories as the underlying computing engine, supporting continuous improvements in communication quality while delivering enhanced customer experiences. Within the partnership, KDDI will contribute commercial network data and lead field trials, KDDI Research will define requirements and develop propagation prediction technology, NVIDIA will provide digital twin infrastructure and accelerated computing platforms, Keysight will supply user equipment emulation technology, and Samsung Research America will contribute virtualized RAN technology.

📊 What This Means (Our Analysis)

This collaboration reflects a practical shift toward creating controlled virtual environments where increasingly sophisticated AI systems can mature before reaching operational networks. By separating AI training from live infrastructure, the participating companies are establishing a workflow that supports repeated experimentation while protecting network stability. The combination of simulation, scalable computing, and commercial network validation creates a structured path for advancing AI-based telecommunications capabilities.

The initiative also demonstrates how specialized technology providers can contribute complementary capabilities toward a shared development platform. Each participant addresses a distinct element, from virtual infrastructure and computing to network simulation, virtualization, and commercial deployment. Together, those contributions create a framework designed to accelerate AI innovation while supporting the long-term evolution of 6G network operations.

📌 Our Take: The success of this collaboration will depend on how effectively virtual intelligence can translate into measurable improvements across future commercial wireless networks.