GSMA Launches Open Telco AI to Bridge Critical AI Gap in Networks

BARCELONA, Spain – March 02, 2026 – The global telecommunications industry has drawn a line in the sand, declaring that general-purpose artificial intelligence is no longer sufficient for its complex needs. In a landmark move, the GSMA today launched Open Telco AI, a global initiative uniting operators, vendors, and researchers to accelerate the development of specialized, 'telco-grade' AI.

The initiative, announced at MWC26 Barcelona, is a direct response to a growing performance gap where even the most advanced frontier AI models struggle to meet the stringent demands of modern telecom networks. Backed by industry heavyweights including AT&T and AMD, Open Telco AI establishes a collaborative framework and a new online portal to provide the shared models, data, and computing power needed to build AI that truly "speaks telco."

Beyond General Intelligence: The Need for 'Telco-Grade' AI

For years, the promise of AI has been to revolutionize network operations, the largest cost center for most telecom operators. Yet, progress has been frustratingly slow. A recent GSMA Intelligence report revealed that a mere 16% of generative AI deployments in the sector have been applied to network operations, a clear sign that existing technology is falling short.

The core of the problem is that general AI models, trained on broad internet data, lack the deep domain expertise required for the telecom environment. They struggle to interpret complex network data, parse dense 3GPP standards documentation, or automate network functions with the near-perfect accuracy required. This "accuracy gap" can be as high as 25%, an unacceptable margin of error in an industry built on "five nines" (99.999%) reliability.

"Today's AI models still fall short of the complexity, precision and reliability the telecom industry demands," said Louis Powell, Director of AI Initiatives at GSMA. "Put simply, AI does not yet speak telco and operators are often deploying technology that cannot meet the required levels of accuracy, safety or efficiency. Establishing clear benchmarks and collaborating across the industry on datasets, models and agentic systems is essential."

Open Telco AI aims to close this gap by fostering the creation of AI systems that possess deep vertical knowledge. This 'telco-grade' AI must understand network architecture, service level agreements, and regulatory constraints to move beyond simple demos and deliver real-world value in mission-critical environments.

A United Front: Collaboration as the Core Strategy

Recognizing that no single company can solve this challenge alone, the Open Telco AI initiative is built on a foundation of unprecedented open collaboration. A vast coalition of over 46 organizations, including operators, vendors, AI developers, and academic institutions, is contributing to the effort.

As founding supporters, AT&T and AMD are making significant initial contributions. AT&T is releasing a family of open telco-specific models that are intentionally hardware and cloud-agnostic, ensuring broad accessibility.

"The telecom industry needs AI that understands the realities of networks – not only generic models repurposed for telco tasks," stated Andy Markus, Chief Data and AI Officer at AT&T. "By contributing our expertise and shaping realistic test environments, we're demonstrating how generative and agentic AI can improve customer experience, reduce operational friction and ultimately create new value."

Addressing the immense computational power required for AI development, AMD is providing critical compute capacity through its advanced GPU platforms and its cloud partner, TensorWave. This contribution lowers a major barrier to entry for many potential innovators.

"Telco networks are among the most demanding and regulated environments for AI and moving from promising demos to telco-grade performance requires an open foundation for data, workloads and compute," said Philip Guido, executive vice president and chief commercial officer at AMD. "Through Open Telco AI, with GSMA and AT&T, AMD delivers the enterprise and AI compute needed to train, fine-tune and run open, telco-grade models efficiently from core to edge."

The initiative is further bolstered by a wide range of partners, including NVIDIA, Orange, SK Telecom, Telefónica, Vodafone, and numerous universities, who are submitting data, models, and use cases to build a rich, shared ecosystem.

Inside the Toolbox: The Open Telco AI Portal

At the heart of the initiative is a new online portal, GSMA.com/open-telco-ai, which serves as a central hub for the co-creation of telco-grade AI. It provides the essential building blocks for developers and researchers, organized around four key pillars.

First, the portal offers access to a growing library of Telco Models. These include the hardware-agnostic models from AT&T, a specialized radio-frequency language model called RFGPT from Khalifa University, and a Large Telco Model (LTM) from AdaptKey AI built on NVIDIA Nemotron.

Second, it provides a repository of Open Data, a crucial element for training effective AI. This includes knowledge graphs, embeddings, and fine-tuning datasets of text, logs, and curated standards material from a host of academic and industry partners.

Third, the portal facilitates access to Compute resources. Through the partnership with AMD and TensorWave, participants can access the GPU power and open toolchains necessary for training and inferencing complex AI models.

Finally, the initiative establishes clear Benchmarks to measure progress. A leaderboard, governed by a "Telco Capability Index," assesses model performance across an initial set of seven telecom-specific tasks. This allows for objective evaluation and encourages a focus on real-world utility rather than general AI metrics. Community programs, such as the AI Telco Troubleshooting Challenge which has already attracted over 1,000 registrants, further stimulate collaborative problem-solving.

The Path to Intelligent Networks: Projecting the Impact

The ultimate goal of Open Telco AI extends far beyond technical benchmarks. The initiative aims to catalyze a fundamental transformation in how telecommunications networks are built, managed, and monetized. By equipping the industry with specialized AI, operators can unlock significant operational efficiencies and cost reductions.

The applications range from predictive maintenance that foresees equipment failures before they cause outages to the automated management of network traffic, ensuring optimal performance and energy consumption. This level of automation is seen as a critical step toward the long-term vision of fully autonomous networks that can self-heal and self-optimize with minimal human intervention.

Beyond cost savings, telco-grade AI promises to enhance the customer experience by enabling faster, more accurate support and personalized service offerings. It also opens the door to new revenue streams. As telcos move beyond using AI for internal efficiencies, they are exploring new enterprise services, such as offering GPU-as-a-service or sophisticated edge AI capabilities, leveraging their vast distributed infrastructure. The Open Telco AI initiative provides the foundational tools to accelerate this strategic shift, paving the way for a new era of intelligent, automated, and more valuable communication networks.