AI has scaled. Telcos must rethink enterprise models.

Insights

• AI has moved from experimentation to enterprise scale, shifting focus from proving value to absorbing AI across the organization.
• While technology has evolved well, legacy operating models are now the primary constraint, with siloed, task-based automation limiting system-level outcomes.
• Real value unlocks when telcos move from task-based execution to intent‑led orchestration.
• Infosys’ Frontier Telco model provides a practical framework, enabling telcos to operate as AI-first enterprises.
• Delaying enterprise redesign risks structural disadvantages in the AI era.

For much of the past few years, the telecom industry has carried energy and enthusiasm for artificial intelligence (AI) experimentation. Pilots and use cases ran in customer care, network operations, IT service desks, and pricing teams. AI showed promise as a productivity lever, shaving hours from workflows by automating repetitive tasks.

That phase feels like it is over, shifting the leadership question from ”does AI work?” to ”how can my organization absorb AI at scale?”. As our previous article on Frontier Telco describes, AI has evolved from being a useful tool layered on top of current infrastructure into being the operating logic that shapes how work happens, how decisions are made, and how value is created. And that is precisely where many telcos are under strain.

The 2026 Mobile World Congress in Barcelona further echoed this sentiment. Conversations with telecom operators, technology partners, and ecosystem players confirmed that telcos have made steady progress toward integrating AI into their network operations. However, that shift has brought with it the understanding that realizing the full value of AI — across both current and future business models — requires a fundamental reimagining of the operating model.

Scaling AI exposes structural problems

Many telcos have until now followed a logical path. They modernized infrastructure, migrated workloads to the cloud, cleaned data, and then layered AI on top. Initially, the results were encouraging with faster ticket resolution, improved fault prediction, and better customer insights.

But at scale, this approach reveals its limits.

Task-based automation fragments business intent. Each function optimizes for its own metrics — mean time to repair, cost per ticket, utilization — without a shared view of outcomes. Automation, done in siloed pockets, accelerates individual steps but it does not improve the system as a whole, which leads to higher decision latency and hence, slower progress.

Leaders at MWC described the consequences of leveraging AI on the sidelines rather than integrating it into their enterprise models:

• AI agents act efficiently in isolation but cause conflicts at the system level.
• Pricing and sales models cannot adapt fast enough to AI-driven demand signals.
• Finance functions struggle to model return on investment (ROI) in a fast-moving environment, with pilots unable to prove scalable impact.
• Operations teams are overwhelmed, managing thousands of models, agents, and control loops.

In short, AI does not fail at scale. The legacy operating models do.

The growing mismatch

As AI deployments expand, leaders are realizing that while technology has evolved rapidly and can offer benefits, their own organizational models haven’t evolved at the same rate, creating constraints that technology alone cannot solve.

1. The first pressure point is architecture. AI workloads increasingly span cloud, private data centers, edge environments, and even end-user devices. Intelligence has shifted from being centralized to being more distributed, dynamic, and highly contextual. Orchestrating this intelligence distribution demands a fundamentally different way of designing systems and workflows.
2. The second pressure point is economics. Scaling AI brings rising inference volumes, higher power consumption, and escalating infrastructure costs. GPU density, data center capacity, and energy availability have all taken center stage during board-level discussions.
3. The third, and the most consequential, pressure point is organizational. Traditional telecom operating models remain optimized for task execution, siloed ownership, and linear handoffs. AI does not operate that way. It works best when it acts on intent, adapts in real time, and coordinates actions across domains.

How AI is forcing a reassessment

To benefit fully from AI, telcos must redesign their business and operating models around intent, not tasks. Rather than asking where to deploy the next AI use case, telecom operators are beginning to shift their focus to questions such as:

• How do we preserve business intent end to end?
• How do we orchestrate value across the customer life cycle?
• How do we scale autonomy without losing control?
• Who owns outcomes when machines act autonomously across domains?
• Where does human authority remain essential, and why?
• How are value, risk, and accountability governed when decisions happen at machine speed?
• How does a telco monetize intelligence, not just connectivity?

These questions point toward a different enterprise architecture, one that treats AI as a foundational capability rather than an add-on. This is where the Infosys’ Frontier Telco framework fits best.

Frontier Telco model reframes enterprise operations

This framework recognizes that autonomy at scale requires coordinated changes across three dimensions:

• Intelligent integrated operations, the transactional layer, where AI handles routine execution and humans focus on oversight, exception handling, and system design.
• Customer value orchestration, the operational layer, where stable, cross-functional teams own outcomes across the full customer journey.
• Strategic stewardship and enablement, the strategic layer, where leadership governs direction, ethics, capital allocation, and risk with AI-assisted insight but human judgment.

Telco operators investing in autonomous networks emphasized the need to embed AI directly into control loops, not just dashboards. But they also stressed guardrails, which are the clear boundaries for machine action, transparency, and human override.

Commercial leaders highlighted that AI-driven sales enablement is accelerating the delivery of customized solutions. Technology leaders spoke about integrated platform approaches — neutralized technology stacks with auditability, agent control towers, and harmonized workflow and data models — that enable end-to-end orchestration. These approaches reduce fragmentation, simplify integration, and allow services to be dynamically composed, rather than stitched together through custom code. And an underlying theme across leaders from different functions was the discussion about people. Leaders acknowledged that AI success depends less on algorithms and more on how roles evolve. Engineers become supervisors of systems. Sales teams shift from product selling to value design. Finance teams model scenarios continuously rather than annually, and in each of these cases, the shift was the same: from managing tasks to orchestrating intent.

The inflection point is now

The telecom industry has seen many technology cycles in the past: 3G, 4G, 5G, cloud, virtualization. AI is different in one crucial respect: it changes how decisions are made, not just how systems perform. That change is irreversible.

AI-native competitors are emerging with structurally different cost curves, faster innovation cycles, and adaptive customer models. Telcos that delay enterprise redesign risk becoming structurally unfit, regardless of network quality.

At MWC, there was little appetite for waiting. Leaders recognized that six-month planning cycles are incompatible with AI’s pace. At the same time, rushing without architectural clarity carries equal risk.

The window for leadership lies in deliberate, enterprise-wide transformation that is grounded in intent, governed responsibly, and executed at scale.

What must telco leaders do next?

A clear set of priorities has emerged for telecom leaders:

• First, reframe AI as enterprise infrastructure. This requires redesigning workflows, decision rights, and accountability around outcomes.
• Second, invest in orchestration over automation. Autonomy emerges when systems coordinate intelligently across domains.
• Third, elevate governance as a value creator. Stewardship of AI — covering ethics, sovereignty, security, and ROI — becomes a competitive advantage when done well.
• Finally, build new roles and capabilities deliberately. AI-native telcos require talent that can balance machine intelligence with human judgment, speed with control, and innovation with trust.

Telcos that continue to pursue incremental AI will see limited gains. Those that redesign their enterprises around intent, autonomy, and stewardship will unlock new sources of value across networks, customers, and ecosystems. The industry is moving toward the frontier telco model and value will increasingly concentrate with those prepared to make that shift.