Artificial intelligence has become a central focus across telecom and digital organizations. Across the sector, investment is increasing in data platforms, advanced analytics, machine learning, generative AI, and large language models. These investments are important. In many cases, they are necessary. But for leadership teams, the more important question is not whether the organization has AI capability. It is whether that capability is translating into measurable commercial impact.

The gap between capability and outcome

In many organizations, AI initiatives are still structured as technology-led programs, innovation tracks, or isolated experiments. These efforts can build important capability. But they do not always translate into the outcomes that matter most to the business:

• Revenue growth
• ARPU improvement
• Churn reduction
• Customer lifetime value
• Sales productivity
• Cost-to-serve reduction

This challenge is becoming more visible with generative AI and LLMs. Many early LLM use cases focus on customer support, internal productivity, content generation, and knowledge management. These are useful applications, and they can improve efficiency. But efficiency alone is not the same as commercial transformation.

What the major telecom operators show

The largest telecom operators are approaching AI from different angles. AT&T has linked AI to operating efficiency, digital migration, and cost reduction, including reporting over $1 billion of cost savings in 2025 and a plan for additional annual cost savings by the end of 2028 as it leverages AI and digital experiences. Verizon is positioning AI partly around infrastructure and enterprise demand, including its AI Connect strategy for low-latency edge computing and AI workloads. T-Mobile has emphasized customer experience transformation, personalization, and digitalization, including reporting a 50% reduction in calls to care since 2021 as it continues to simplify and personalize customer experiences through digitalization and AI. These examples are useful because they show that AI is not one thing. It can be an efficiency lever, a network/infrastructure lever, a customer experience lever, or a commercial growth lever. The strategic question is: which problem is the business trying to solve first?

A more commercially anchored approach

The most effective AI applications start with a business problem, not a technology capability. The starting point should shift from: "What can we build with AI?" to: "Where can better decision-making improve commercial outcomes?" In telecom, the strongest commercial use cases usually sit around:

• Customer acquisition
• Retention and churn management
• Upsell and cross-sell
• Pricing and offer personalization
• Lifetime value optimization
• Sales and service enablement

This is where AI becomes more than an innovation initiative. It becomes part of the commercial operating model.

Where LLMs can add real value

LLMs are often associated with chatbots and productivity tools, but their broader commercial potential is more significant. In telecom, LLMs can create value when applied to areas such as: Customer interaction intelligence Understanding intent, sentiment, complaints, and unmet needs from customer conversations in real time. Offer and pricing personalization Supporting more relevant messaging, bundles, and recommendations based on customer context. Sales and retention enablement Helping frontline teams understand the customer situation and identify the next best action. Operationalizing unstructured data Extracting insight from complaints, call notes, chats, surveys, social feedback, and service interactions. The value is not in the LLM alone. The value comes when these outputs are connected to commercial decisioning frameworks.

A practical example

In one market, we implemented AI-driven recommendation models focused on usage behavior, consumption patterns, and customer segmentation. The objective was not experimentation. The objective was commercial impact. The models were embedded into marketing campaigns, offer design, and customer engagement channels. This allowed the business to target customers more effectively, improve conversion, and strengthen retention. The key was not simply having the model. The key was integrating the model into the way the business made decisions.

Integration matters more than capability

One of the most important lessons from AI implementation is that value rarely comes from the model by itself. Value comes from how the output is used. That requires:

• Embedding insights into day-to-day commercial decisions
• Aligning marketing, sales, service, data, and technology teams
• Creating feedback loops to refine performance
• Measuring business outcomes, not only technical performance

Without this integration, even advanced AI and LLM capabilities remain underutilized.

What executives should govern

At executive level, AI should be governed with the same discipline as any other strategic investment. That means leadership teams should be clear on:

• Which commercial or operational problem the use case solves
• Which KPI it is expected to influence
• Who owns the business outcome
• How the insight will be operationalized
• How performance will be measured and refined

AI should not be evaluated only by model accuracy, deployment speed, or number of use cases. It should be evaluated by its contribution to business performance.

Final reflection

AI and LLMs have significant potential to improve how telecom businesses operate, compete, and grow. But their value is most clearly realized when they are anchored in commercial priorities, embedded into operational workflows, and governed with financial discipline. In that context, AI becomes less about innovation for its own sake. It becomes a way to enable better, faster, and more consistent commercial decisions. The real question is no longer whether telecom operators are investing in AI. Most are. The question is whether those investments are changing the decisions that drive revenue, retention, customer value, and performance.