Scaling Telco AI from Ambition to Reality

GSMA's Louis Powell and TM Forum's Andy Tiller highlight a new collaboration to help the industry reduce fragmentation and accelerate practical telco AI adoption.

Artificial intelligence is now a board-level priority across the telecoms industry. Operators are investing, vendors are innovating and the wider ecosystem is moving quickly. Yet, despite this momentum, telecoms continues to face structural barriers that make AI deployment far more complex than in many other industries.

Telecom is one of the hardest environments for AI to transform. Networks are multi-vendor, highly fragmented, and reliant on siloed data, with very little tolerance for error. To be useful in this context, AI must be accurate, efficient, and trusted from day one.

Today's frontier models are not designed for telecom. They are not trained on telecom-specific data and often struggle with the language, architecture, and operating realities of networks. As a result, deployments remain concentrated in customer experience and enterprise functions, rather than in the network layer itself. GSMA Intelligence data shows that just 16% of AI deployments target network use cases, despite networks accounting for 34% of OPEX as cost pressures and expectations for return on investment continue to rise.

That is why telecom needs to collaborate and align to solve these challenges for itself, building AI approaches that reflect the complexity, economics, and operational realities of the industry.

Closer Collaboration between GSMA and TM Forum

The shared challenge is clear: telecoms needs AI that understands telecoms, and a clearer path from experimentation to deployment at scale.

This is where deeper collaboration between GSMA and TM Forum adds real value, addressing complementary, parts of the same challenge.

Through the Open Telco AI initiative, GSMA is focused on building telco-grade AI; models, and agentic systems that are accurate, efficient and trusted. That means creating a shared foundation of telco models trained on specialised telecom datasets, such as the OTEL family of models, and benchmarks that demonstrate how these models perform on real telecom tasks.

TM Forum is focused on frameworks, standards and high-value use cases that enable industry alignment on what should be built, how it should interoperate and where it should create value first. Its Autonomous Networks and Trustworthy AI and Data Missions create frameworks for scaling AI within its Open Digital Architecture (ODA), alongside solution packs for high-value scenarios and standardised tools for measuring progress. Catalyst and Innovation Hub projects build proofs of concept and reference implementations for deploying these solutions.

This is about connecting the AI stack end-to-end, combining shared models, common frameworks and practical deployment pathways, with the trust required for critical network environments.

Combined approach already making an impact

The aim of closer collaboration is to reduce fragmentation and accelerate industry-wide adoption by bringing these elements together in a more coherent way. The impact is already visible.

Philippine mobile network operator, Globe Telecom, is exploring multi-vendor RAN root cause analysis using TM Forum's technical solution packs to help expose normalised APIs across vendors, while also working with Open Telco AI on the models and agents needed to automate analysis.

Similarly, AT&T's OTEL models, specialized models for telco, are being adapted into TM Forum environments such as Model-as-a-Service (MODaaS) and ODA Canvas, transforming proven, operator-grade AI into standardised, interoperable assets that the wider telecom industry can adopt and scale across multi-vendor, cloud-native environments.

These examples matter because they tangibly demonstrate end-to-end alignment, from use case definition to model development to deployment. It shows how AI can move beyond isolated innovation, towards interoperable AI models that are designed to operate within established industry frameworks. And it is a collaborative ecosystem in which operators, vendors and partners can help shape shared solutions rather than navigating disconnected efforts in parallel.

Telecoms sits at the heart of the AI economy.

Networks underpin the cloud, device and edge capabilities on which AI depends. Yet telcos themselves have not fully captured the value of AI within their own operations.

If the industry is to accelerate towards more autonomous, AI-native networks, it must align on shared assets, common standards and practical implementation pathways.

That is why this collaboration matters now. the industry cannot rely on generic tools and fragmented progress. It needs a shared, trusted foundation. That means engaging in Open Telco AI and TM Forum initiatives, contributing to use case definition, model development and validation, and helping shape the standards, frameworks and architectures that will determine whether AI delivers real value in telecom networks.

The opportunity is significant. But unlocking it requires alignment as well as ambition. GSMA and TM Forum are committed to delivering both, and this is just the start, with more to come throughout the year on data, models, evaluations and Member proof points.

Look out for more on telco AI at TM Forum and GSMA events this week.

DTW Ignite in Copenhagen opens its doors today, where TM Forum will spotlight trustworthy AI, autonomous networks and real-world Member showcases that demonstrate how AI-native telecoms can move from concept to execution.

And at MWC Shanghai (24-26 June), GSMA will bring together industry innovators on the GSMA Pavilion and through Open Telco AI sessions focused on building telco-grade AI benchmarks and practical deployment pathways.