Telco Giants Launch AI Alliance
The GSMA, the mobile industry's main trade group, has launched a new initiative called "Open Telco AI." The project aims to standardize and accelerate the use of AI for network operators worldwide, reflecting a major push to embed artificial intelligence deeper into global communications infrastructure.
The push for this alliance stems from a significant performance gap; general-purpose AI models often struggle to accurately interpret complex network data and automate operations. This has limited progress, with only 16% of generative AI deployments in the telecom sector being applied to network operations. The GSMA's Director of AI Initiatives, Louis Powell, noted that current AI "does not yet speak telco." Founding supporters AT&T and AMD are making initial key contributions. AT&T is releasing a set of open telco-models trained on public data to be hardware-agnostic, while AMD is providing the necessary GPU platform computing power for model training and evaluation. The initiative will establish a "Telco Capability Index" to measure how well AI models perform on telecom-specific tasks. This open, collaborative approach aims to create a shared foundation and clear benchmarks for what constitutes "telco-grade" AI. This effort is distinct from the Global Telco AI Alliance (GTAA), formed in July 2023 by operators like Deutsche Telekom and SK Telecom, which is focused on creating a multilingual Large Language Model (LLM) primarily for improving customer service chatbots. The Open Telco AI initiative has a broader focus on network operations. A wide range of global operators and tech firms have joined as participants and partners, including China Mobile, Deutsche Telekom, Google Cloud, IBM, Nvidia, Telefónica, and Vodafone. The collaboration aims to solve challenges like vendor-specific data formats, which complicate efforts to get a unified view of a network. By creating knowledge graphs and facilitating data model translation, generative AI can help harmonize data across diverse systems. Ultimately, embedding more sophisticated AI into network operations could reduce total network operating expenses by 15% to 30%. This would be achieved through predictive maintenance, dynamic energy consumption management, and automating the resolution of network issues, sometimes before customers are even aware of a problem.
