Research Report
The network that thinks for every customer
Building autonomous networks for personalized telecom experiences
5-minute read
August 26, 2026
In brief
- As AI reshapes customer expectations, telcos need networks that know each customer and act before a problem is noticed: a network of one.
- Autonomous networks aren't just a fix for reliability and cost efficiencies. By 2030, network operators expect 11% higher revenues and 30% lower operating expenditure.
- Just 13% have an enterprise-wide network strategy; only 18% have CEO-led AI network transformation initiatives. Telcos need both to build autonomous networks.
Most networks weren’t built for what customers need now
The call that never drops. Live streaming that never buffers. The factory floor that never stalls. The delivery that never gets lost.
From consumers to enterprises, across all sectors telcos serve, customers aren't buying network performance, they're buying dependable outcomes. The new expectation is simple and unforgiving: known, anticipated, served before they ask—and the network that can't deliver that is already behind.
Telecoms are judged on connectivity working when it matters most
Right now, many telco companies struggle to consistently meet people’s expectations. In earlier customer–focused research we found:
67%
of people rank reliable network quality as the most important factor when choosing a telecom provider.1
64%
of people cite service outages as one of their biggest frustrations.2
40%
of enterprise customers are dissatisfied with connectivity services, citing network reliability and real-time threat detection as key concerns.3
For years, speed, latency and coverage set high-performing telcos apart. Today, exceptional performance is the baseline, reliability is assumed, and neither is enough. Customers shaped by AI that remembers and anticipates them now bring that same expectation to their network. The telco that gets this right becomes something more than a network provider. It can deliver a “network of one”: a tailored customer experience resulting from autonomous capabilities that monitor customer requirements, remember context, anticipate needs and prevent disruption.
For the leaders who act, the network is poised to become their most differentiated asset, the one thing a competitor cannot replicate. When the network becomes that differentiated, the commercial model changes with it.
Meeting people’s expectations is getting harder, not easier
AI workloads, edge applications and streaming traffic are making networks harder to manage. Yet many operators are still running on infrastructure built for a simpler time—systems that don’t share data, don’t coordinate decisions, and weren’t designed for the pace and complexity they are now asked to handle. The result is operators having to constantly react to problems their systems should have autonomously predicted and handled. In our research with 250 senior telecom executives across the globe:
1 in 4
telcos report end-to-end visibility across network domains and lifecycle stages.
26%
of operators believe their incident response processes rapidly isolate and recover from outages.
1 in 5
telcos extensively use AI, analytics or automation to proactively prevent disruptions.
Customers are not becoming more demanding. They are simply applying the standard they already experience everywhere else. For a network still built around response, that distinction is the whole problem.
Telcos know the destination, but scaling autonomy is a challenge
Autonomous networks detect, decide and act, all before customers notice anything is wrong. But their value goes beyond preventing disruption. By continuously optimizing performance, security and service delivery, these networks help operators to lower costs, improve customer outcomes and give telcos a platform to launch the next generation of services for customers and businesses.
Telco executives see the potential. Network budgets allocated to AI are expected to nearly double, from 9% today up to 17% by 2028. And the financial case is clear. By 2030, operators who successfully scale autonomous networks expect to deliver:
30%
lower network operating expenditure.
17%
lower network capital expenditure.
11%
higher revenue.
This confidence has yet to translate into capability. In most telco companies, automation is confined to specific tasks or processes and a large share of network diagnosis, decision-making and execution is manual. They remain in the intermediate stages on the six-level network autonomy scale, from level zero, where operations are fully manual, to level five, where networks can manage themselves with minimal human intervention. Only one in three telcos expect to achieve conditional to highly autonomous networks (level four to five) by 2030.
Given how fast agentic AI has moved, if we asked those 250 executives today, they’d likely be more optimistic. But technology alone doesn’t create autonomous networks; operators still need their foundations and organization in place to support them. The same is true of personalization: the right data and architecture let telcos build a real-time relationship with each subscriber, offer a network of one, anticipating needs before issues surface. The infrastructure stays the same. What changes is the data, its source, its richness and the agentic architecture that puts it to work.
Many telcos haven’t landed this yet.
Three barriers hold telcos back
While executives are confident in the outcomes and agentic AI is accelerating what’s possible for autonomous networks faster than the industry anticipates, three structural barriers stand between telcos and truly autonomous networks.
01
AI transformation ownership and strategy are fragmented
Autonomous networks cannot be built by optimizing individual network domains. They require CEO-led, enterprise-wide transformation.
Yet only 18% of those surveyed say they have CEO-led AI network transformation initiatives, and just 13% have an enterprise-wide autonomous network strategy.
Without clear ownership and a shared strategy, efforts and impact remain localized and limited.
DATA
18%
of telcos have CEO-led AI network transformations
02
Data and talent readiness remain key constraints
Autonomy depends on high-quality data and new skills. 81% of telecom executives cite data quality as the biggest challenge to scaling autonomous networks. At the same time, 77% say hiring AI talent is significantly difficult, and 51% face challenges reskilling their existing workforce. Telcos are being asked to build intelligence-led networks with foundations that weren’t designed for intelligence.
These barriers don’t just slow the path to reliability and resilience. They’re blocking the move to an entirely different commercial model. Autonomous networks make a new generation of advanced, high-value services possible and they make the previous commercial model obsolete. Autonomous decisions are made on fragments, and fragmented decisions are not intelligence. Fail to clear these barriers and telcos aren’t just running behind on customer experience. They are leaving the next wave of growth on the table.
Data
81%
of telco execs say data quality is the biggest challenge
03
Operating models have not evolved for AI
Embedding AI into existing workflows can optimize individual tasks, but it cannot transform how the network is designed, run and managed end to end. Only 17% of telcos are currently pursuing the fundamental redesign of network processes that level four/five autonomy requires.
That leaves most operators trying to evolve the old model rather than building the new one. Without end-to-end redesign, AI remains a point solution, not a transformation.
Data
17%
of telcos are redesigning network processes for AI
How leaders can scale autonomy to drive resilience and growth
None of this is out of reach. Working with our clients, we see the operators pulling ahead share a pattern and it starts smaller than most expect.
They begin by building a shared network intelligence layer that connects data, AI models, agents and orchestration into a single architectural core. Central to this is what we call a Digital Brain that senses conditions, reasons across domains, orchestrates the best response, learns from outcomes and acts autonomously throughout the network lifecycle.
The leaders don’t try to automate everything at once. They start where the payoff is clearest: for example, in the network operations center (NOC), where cost, complexity and manual effort are concentrated. Early wins establish the business case and build the organizational momentum to go further.
Crucially, leading telcos treat autonomous networks as an enterprise transformation rather than an IT initiative. They reinforce this with CEO sponsorship and shared objectives tied to customer and business outcomes. They stop running the network as separate domains each with its own teams and targets. Instead, they design operating models around connected decisions across the full network lifecycle. And they build the foundations deliberately: data, governance and workforce are shaped from the outset for an AI-first way of working.
From operator to orchestrator
The goal is to connect decisions and actions across the whole network to build intelligent, self-managing networks that anticipate the needs of every customer.
These foundations enable autonomy at scale and position operators to monetize network intelligence through new AI-enabled services and business models for both consumers and businesses. The operators who move first won’t just run more resilient networks. They’ll be the ones capturing a larger role in the emerging AI value chain.
The real breakthrough: moving beyond operating one shared network for millions of customers to a network of one. When customer service becomes the primary driver of autonomous networks, the network is primed to become the operator’s most valuable and differentiated asset.
Sources:
Unless stated, all data is from Accenture Communications Industry Network Research Executive Survey, January-February 2026, N=250.
1, 2: Accenture, Empowering Loyalty: How AI can transform CSPs B2C growth, June 2025, n=6,800
3: Reinventing B2B for the AI economy, November 2025, n=1,200