The telecommunications sector stands at a critical juncture. While Artificial Intelligence (AI) has become a boardroom imperative across every major industry, the path to AI-driven network operations for telcos remains fraught with technical hurdles. Recognizing this, a strategic collaboration between the GSMA—the global organization representing mobile operators—and the TM Forum—the industry authority on digital transformation standards—has been launched to harmonize the industry’s approach to AI deployment.
Led by the GSMA’s Louis Powell and TM Forum’s Andy Tiller, this partnership seeks to dismantle the structural fragmentation that has long plagued network-layer AI, aiming to accelerate the industry’s transition toward autonomous, AI-native networks.
1. The Core Challenge: Why Telecom is Different
The excitement surrounding Large Language Models (LLMs) and generative AI has often overshadowed the realities of deploying these technologies in highly regulated, mission-critical infrastructure. Unlike the consumer tech sector, telecommunications operates on multi-vendor, heterogeneous environments where downtime is unacceptable and data silos are the norm.
The Accuracy and Trust Gap
Frontier AI models, trained primarily on general internet data, often lack the specialized "domain knowledge" required to manage network architecture. They struggle with the specific nuances of signaling protocols, operational language, and the precise, error-intolerant nature of network maintenance. Consequently, AI adoption has been largely relegated to "safe" customer-facing functions—such as chatbots and billing queries—leaving the core network, which accounts for nearly 34% of typical operator OPEX, largely untouched.
According to data from GSMA Intelligence, only 16% of AI deployments currently target network use cases. This represents a significant missed opportunity for cost optimization and operational efficiency. To bridge this gap, the industry requires AI that is not just "general-purpose," but "telco-grade."
2. Chronology of the Collaboration
The push for this collaboration did not happen in a vacuum. It is the culmination of years of siloed development and the recent explosion of generative AI, which highlighted the necessity for a unified industry stance.
- Pre-2023: Individual operators began exploring AI for network maintenance, largely relying on proprietary, vendor-locked solutions that lacked interoperability.
- Early 2024: The GSMA formally intensified its focus on "Open Telco AI," aiming to build a shared foundation of specialized models trained on industry-specific datasets, such as the OTEL family.
- Mid-2024: TM Forum accelerated its work on Autonomous Networks and the Open Digital Architecture (ODA), providing the structural frameworks needed to support AI at scale.
- Late 2024/Early 2025: The realization that model development (GSMA) and framework standardization (TM Forum) were two sides of the same coin led to the formalization of this collaborative effort.
- Present Day: The alliance is now focusing on "end-to-end" alignment, moving from theoretical proofs of concept to standardized, deployable assets for global operators.
3. Supporting Data: The Economic Imperative
The financial argument for this collaboration is compelling. With networks consuming over a third of an operator’s operating budget, even minor improvements in automated fault detection or traffic management yield massive bottom-line benefits.
- The 34% Factor: Networks remain the largest cost center for telcos. By moving AI from the customer service desk to the network core, operators can target the highest potential for ROI.
- The 16% Benchmark: The current low penetration of network-specific AI underscores the "complexity tax" that prevents adoption.
- Multi-Vendor Reality: With most operators relying on a mix of vendors (Ericsson, Nokia, Huawei, etc.), the need for common, normalized APIs—which the TM Forum is facilitating—is the only way to avoid vendor lock-in.
4. Official Perspectives: A Synergy of Missions
The collaboration is structured to ensure that the GSMA’s model development and the TM Forum’s architectural standards function as a cohesive stack.
GSMA: The "Open Telco AI" Engine
Louis Powell, representing the GSMA, emphasizes that the goal is to build an ecosystem of "accurate, efficient, and trusted" systems. The Open Telco AI initiative is designed to create a shared repository of telco-specialized models. By training models on telecom-specific datasets, the GSMA aims to ensure that AI can accurately interpret network logs and provide actionable insights that generic frontier models would miss.
TM Forum: The Architectural Framework
Andy Tiller of the TM Forum argues that having a great model is insufficient if it cannot be integrated into existing OSS/BSS (Operations Support Systems/Business Support Systems) environments. The TM Forum provides the "connective tissue." Through its Open Digital Architecture (ODA) and "Model-as-a-Service" (MODaaS) framework, the organization is creating the standard plumbing required to ensure that AI models can talk to diverse network components, regardless of the vendor.
5. Real-World Impact: From Concept to Execution
The efficacy of this partnership is already being tested in the field, moving beyond white papers and into live network environments.
The Globe Telecom Case Study
In the Philippines, Globe Telecom is currently navigating the complexities of multi-vendor Radio Access Network (RAN) root cause analysis. By utilizing TM Forum’s technical solution packs, Globe is standardizing its APIs to create a normalized data layer. Simultaneously, they are integrating Open Telco AI models to automate the analysis of these data streams. This two-pronged approach allows the operator to identify network faults faster and with greater accuracy than traditional, rule-based systems.
The AT&T and MODaaS Integration
AT&T’s work with the OTEL family of models is serving as a blueprint for the industry. By adapting these specialized models into the TM Forum’s ODA Canvas, AT&T is demonstrating how "operator-grade" AI can be packaged into interoperable assets. This means that an innovation developed in one part of the world can, in theory, be deployed in another, reducing the need for costly, redundant development.
6. Strategic Implications for the Industry
The success of this collaboration will likely dictate the winners and losers in the next decade of telecommunications.
Moving Toward Autonomous Networks
The ultimate goal is the "Autonomous Network"—a self-healing, self-optimizing infrastructure that requires minimal human intervention. To achieve this, the industry must move away from the current state of "isolated innovation," where operators build bespoke solutions that cannot scale. The alignment between GSMA and TM Forum forces vendors to build to common standards, which in turn reduces the barrier to entry for smaller, more agile AI providers.
The Shift in Power Dynamics
As networks become "AI-native," the power dynamic between operators and equipment vendors is shifting. Standardized frameworks mean that operators have more leverage to demand interoperable solutions. This fosters a more competitive ecosystem where vendors compete on the quality of their AI-driven features rather than their ability to lock an operator into a proprietary ecosystem.
A Call to Action
The leaders of this initiative are clear: the time for individual, fragmented experimentation has passed. They are calling on the wider industry to:
- Engage in Open Telco AI: Contribute data and use cases to help train the next generation of industry models.
- Adopt TM Forum Standards: Align internal architectures with the ODA to ensure that new AI tools are "plug-and-play" ready.
- Validate and Benchmark: Use the industry-standard benchmarks being developed to track progress and prove the value of AI in the network.
7. Conclusion: The Path Forward
The collaboration between the GSMA and TM Forum is more than just a partnership; it is a necessary evolution for a sector that has historically been siloed. By combining the GSMA’s strength in model development with the TM Forum’s mastery of architectural standards, the industry is finally building the bridge between AI potential and operational reality.
As the industry gathers at DTW Ignite in Copenhagen and prepares for MWC Shanghai, the focus is clear: moving from the conceptual promise of AI to the tangible, scalable execution of AI-native networks. The roadmap for 2025 and beyond is set, and for the global telecommunications sector, the message is one of alignment, shared assets, and a unified push toward an intelligent future. The infrastructure that supports the global AI economy is finally preparing to use that very same intelligence to secure its own survival and growth.
