Beyond Automation: Agentic AI as the
Catalyst for Unlocking Telco Growth

ORDER REPRINTS DOWNLOAD COMMENT DISCUSS SHARE

Autonomy is coming in a new way for Communications Service Providers (CSPs) with the arrival of agentic AI. Unlike previous waves of automation, this phase introduces lightweight, task-specific, and even ephemeral agents capable of collaborative action and reasoning beyond the constraints of hardwired workflows. These AI systems move the industry from ‘human in the loop’ to ‘human in command,’ where agents act independently towards goals while humans oversee and manage outcomes. The convergence of advanced compute, disruptive storage, new data innovation practices, and neuroscience-inspired architectures marks a significant leap – enabling CSPs to unlock horizontal scaling, compressed time-to-value, and richer digital transformation than ever before.

All the ingredients, but what’s the recipe?

CSPs are uniquely positioned to unlock the full potential of agentic AI. They already possess the essential ingredients for success: mature infrastructure, advanced technologies, and skilled talent. As data-rich organizations by design, CSPs operate across highly distributed architectures and adhere to some of the strictest standards of regulatory compliance, sovereignty, and security – traits that are increasingly critical for deploying trusted AI at scale. Their operational track record managing complex systems enables them to pursue AI-driven transformation with confidence and credibility.

This foundation not only supports resilient and scalable AI deployments but also empowers CSPs to shift from traditional service delivery to new, revenue-generating opportunities – such as MLOps-as-a-Service, advanced security solutions, and high-value data products. In short, CSPs aren’t just ready for agentic AI – they’re built for it.

Yet despite having all the ingredients, we’re still looking for the recipe. Despite major investment, the success rate of AI initiatives remains disappointing, with more than 50% of AI initiatives by CSPs failing to scale or reach production. Use cases often stay stuck in experimentation, and the anticipated transformation falls short. Projects struggle to transition from isolated pilots to business-critical production, and the ROI of AI, especially for productivity-oriented projects, is hard to quantify. At the same time, many deployments become siloed, compounding technical and data debt rather than solving for industry-wide growth.

Three dimensions for success

To truly capitalize on AI’s potential, CSPs must overcome structural, operational, and cultural blockers that limit scale and impact. At TM Forum, we believe this challenge can be broken down into three essential dimensions – each requiring intentional action and alignment across business and technology. Success isn’t just about choosing the right tools; it’s about reshaping how leadership leads, how teams deliver, and how technology is adopted and governed across the organization.

1. AI for Leadership: Bridging the business-technology divide
AI cannot thrive in a vacuum. Too often, promising initiatives stall because leaders lack the frameworks to articulate clear, ROI-driven business cases or to plan AI transformation in a structured way. The result is fragmented pilot projects that rarely scale or deliver strategic value. Leadership must step up – not just to sponsor AI, but to own it. This means equipping C-level and business leaders with a pragmatic language for AI that translates ambition into outcomes, and vision into measurable value.

Without that executive clarity and cross-functional alignment, AI remains trapped in isolated silos instead of becoming a muscle for growth.

2. AI for Everyone: Breaking through delivery inhibitors
Even with leadership support, AI delivery often hits systemic roadblocks. Programs can be difficult to set up and frequently lose momentum before producing real results. Financial justification is another hurdle – AI’s productivity gains are often indirect or distributed, making ROI hard to pin down. On the technical front, many teams lack realistic ways to estimate AI’s architectural impact or secure operational buy-in for new capabilities.

Add to this the absence of standardized delivery accelerators – like reusable tools, templates, and proven playbooks – and it’s easy to see why progress can feel painfully slow. To scale AI, CSPs need to democratize delivery with a clear path from experimentation to industrialization.

3. AI-First Technology Adoption: Empowering sustainable transformation
At the heart of scalable AI-native operations lies a set of foundational capabilities that many CSPs are still developing. It starts with data: rather than attempting to wrangle