Turning telecom AI investment into growth

Top telecom leaders recently met at Mobile World Congress (MWC) in Barcelona to discuss how AI is transforming telecom. The key takeaway? AI is disrupting the entire tech stack, but more efficient operations is just the starting point.

Growth is the main priority in telecom. Leaders want to unlock the next billion euros in revenue using innovations such as AI agents. But they say there’s a gap between AI benefits in theory and practice.

After massive spending on 5G, network modernisation, and AI initiatives, telecom operators want to see a clear connection between investments and returns. Let’s unpack how to turn telecom AI investment into value.

Where’s the revenue from telecom AI?

According to the 2025 ServiceNow Enterprise AI Maturity Index, more than half (55%) of global organisations have implemented at least 100 AI use cases. Most of them (82%) expect to increase their AI investments in the next year. So rather than asking, “Are you using AI?” telecom leaders are instead asking, “Where’s the revenue from AI?”

Our research found that global AI maturity declined by nine index points in the past year. It’s not because AI isn’t powerful—organisations using AI report improved experiences, increased productivity, and faster innovation. The decline is a result of many businesses layering AI on broken processes.

This dynamic can block organisations from achieving growth. To achieve value, organisations need a strong foundation that connects data, AI, and workflows—particularly in telecom, which has some of the most complex architectures of any sector.

AI can disrupt the telecom tech stack

AI implemented across the tech stack can deliver comprehensive transformation. Let’s assess the impact from the top down.

AI is changing how customers and employees interact with digital systems. Multi-modal AI experiences—which allow users to communicate using any combination of voice, text, or images—are becoming the new way to engage with software. By implementing multi-modal experiences, telecom organisations can enhance how they run customer care, field operations, and retail channels.

Organisations are starting to use reasoning models to support automation. Coordinated teams of AI agents can work together to carry out complex, multi-step telecom workflows that weren’t previously automatable—such as order management, fault resolution, and network change orchestration.

The AI layer is evolving quickly. As training costs fall, leaders can implement specialised models that understand network data, customer behaviour patterns, and context from business and operations support systems. These can be combined with agentic models that take action.

It’s not enough for AI to access disparate data sources. Communications service providers need to provide an extra layer of context that connects data across the organisation. This “semantic layer” enables AI to understand relationships between network nodes, billing events, and field tickets so that it can reason and execute processes across the business.

The infrastructure the tech stack is built on must be powerful. The tandem development of chips and models is creating an opportunity for tangible cost and performance benefits.

4 principles for AI agents in telecom

Most AI agents can answer questions. The differentiator for high-performing AI agents is the ability to consistently execute work at scale across complex operations. At ServiceNow, we’re solving that challenge with our telecom partners. It has four parts:

1. Sense

AI models are traditionally trained on publicly available data from the internet, but this provides only general knowledge. Specialised AI models for telecom must understand the relationships between different aspects of the business.

This requires an orchestration layer that connects different systems and equips AI with live, cross-business context. Greater visibility unlocks new use cases. For example, if a network fault arises, AI agents can identify the affected customers and flag the service-level agreements (SLAs) at risk of breach.

2. Decide

AI models that reason well in enclosed training environments must also do so in complex, regulated telecom environments. Any decisions they make should be grounded in both business policy and regulation so that models are auditable and trusted to act appropriately in real-life scenarios.

3. Act

Surfacing an answer inside a tool is a baseline use case for AI in telecom. AI agents must also be trained to execute complex workflows, such as:

• Rerouting traffic
• Triggering a field dispatch
• Escalating an SLA breach

This enables AI agents to move workflows forward with minimal human intervention.

4. Govern

Every AI implementation needs guardrails so that actions are visible, auditable, and compliant. This is particularly important for the telecom industry, as proposed frameworks such as the Digital Networks Act introduce more comprehensive standards for infrastructure resilience.

Telecom AI agents in action

Telecom success depends on customer relationships. When networks don’t work seamlessly, customers may become frustrated and escalate their problems.

One network provider at MWC shared how its escalation team used to lose hours of productivity chasing and validating customer information, flagging duplicate tickets, and manually routing cases to the correct desks.

AI agents now handle the majority of case intake and creation. By the time a case reaches an escalation manager, it’s clean, complete, and already routed. AI helps finish the work so people can focus on human conversations and protecting revenue that might otherwise be at risk.

Find out how ServiceNow can help you put AI to work for telecom.