1FINITY to launch software linking RAN and AI in December

The Nikkei reported that Fujitsu-affiliated 1FINITY announced on November 4, 2026, software with functions that link RAN and AI, and will begin selling it in December. The company said the platform, designed to work with NVIDIA GPU infrastructure and other systems, will allow telecom operators to run base station functions, generative AI and AI agents on the same server. Denpa Times reported the product name as “Open vRAN Solutions VB5000.”

1FINITY says its platform will enable base station functions, generative AI and AI agents to run on the same server, helping improve the efficiency of capital investment. For telecom operators weighing this approach against building separate infrastructure for communications and AI processing, a key condition will be whether they can operate AI services on a shared platform while maintaining communications quality.

The focus of this move is not just the virtualization of base station functions, but a configuration that places RAN and AI processing on a shared computing platform. It extends the existing vRAN trend of running base station functions as software on servers by integrating AI processing. The novelty is not the slogan of adding AI to communications networks, but offering a product that provides a platform for wireless communications and AI to share computing resources. In this configuration, communications and AI processing loads converge on the same computing platform. Whether sharing equipment improves utilization is likely to depend on operating conditions, including how processing is allocated, how performance is maintained and whether communications can continue during failures. 1FINITY is reported to expect improved capital investment efficiency, but assessing the benefits of sharing will require checking whether communications quality can be maintained even when AI service loads fluctuate. The start of software sales alone does not show the operational benefits or scope of application. Compared with KDDI’s physical AI strategy announcement, which we covered on October 27, 2026, this move can be seen as software for handling the communications network itself and AI processing on a shared platform, while the earlier development concerned a telecom operator’s efforts to commercialize physical AI. However, there is no information that the two are collaborating, and they cannot be treated as part of the same business plan. For telecom operators, a key issue will be how far to integrate the infrastructure for providing AI services with the operating infrastructure for wireless networks. The reliance on NVIDIA GPU infrastructure also indicates that product value depends not only on communications equipment software but on how it is combined with computing infrastructure. Going forward, it will be important to see which communications and AI processes are actually run together, how GPU resources and communications quality are managed, and how equipment utilization changes after deployment. The reports confirm the start of sales and the platform concept, but do not provide these operating conditions or results.