The GSMA Foundry and Khalifa University have launched a major strategic collaboration at MWC Doha to accelerate telecom-focused AI innovation. The partnership aims to build specialised AI models, datasets and benchmarking tools tailored to the needs of mobile operators as the industry moves toward more autonomous, intelligence-driven networks.
Under the agreement, both organisations will co-develop advanced AI assets that address telecom-specific challenges—an area where general-purpose large language models still struggle. The first phase includes the release of TelecomGPT, a telco-first large language model with a dedicated chat interface hosted on LightOn, and the Open Telco Knowledge Graph built on 3GPP documentation and hosted on Hugging Face. These tools will support accurate interpretation of standards, troubleshooting, and complex network operations.
GSMA Open-Telco LLM Benchmark results have shown that even the most advanced general LLMs fail to reliably handle specialised telecom queries, highlighting the need for domain-trained AI. The collaboration with Khalifa University’s 6G Research Centre (6GRC) is designed to close this gap by enabling robust, safe, energy-efficient AI systems specifically optimised for telecom workloads.
Khalifa University’s Provost, Professor Bayan Sharif, said the combined expertise of GSMA and the university will drive meaningful innovation in telecom AI, building solutions that overcome critical limitations of current models. With strong research foundations and real-world industry experience, both partners aim to accelerate the development of next-generation AI capabilities for the global mobile ecosystem.
