Telecom Giants Launch 'Open Telco AI'

The "Open Telco AI" initiative was launched to address the significant performance gap of general-purpose AI models in telecommunications. Many existing AI systems struggle to accurately interpret network data or understand specific industry standards, a problem the GSMA says has limited the application of generative AI in network operations to just 16% of deployments. AT&T and AMD are key founding supporters of this global alliance. AT&T is contributing by releasing a family of open-source telco models, while AMD is providing the necessary computing power for training and evaluating these AI systems through its GPU platforms. This collaboration extends to over two dozen organizations, including major operators like Orange, Swisscom, and Vodafone, alongside tech giants such as Google Cloud, IBM, and Nvidia. The initiative also involves academic institutions like Khalifa University and Purdue University to foster broader innovation. The project will establish a "Telco Capability Index" to measure how well AI models perform on telecom-specific tasks. This will create a benchmark for developing more specialized and effective AI for tasks like network troubleshooting and optimization. A key goal is to create a library of knowledge graphs and datasets specifically for fine-tuning AI. This will include curated standards material from the GSMA, logs, and text to help the AI "speak telco" and understand the industry's complex operations. The initiative will provide a portal for open models, data, and tools to accelerate development. This includes specialized models like a radio-frequency language model from Khalifa University called RFGPT and a Large Telco Model from AdaptKey AI built on Nvidia Nemotron.