Spend time at a telco transformation conference and one single word dominates every slide deck: automation. By focusing on building networks that self-configure, self-heal and self-optimise, with the slow hand of human intervention taken out of the routine of operations altogether, zero-touch network automation is no longer a futurist concept. It is, according to TM Forum’s autonomous network maturity framework, an industry North Star that telecoms operators are already racing towards.
But there is an uncomfortable truth that too few operators are willing to say plainly that while the technology is largely ready, the processes and culture are not.
Automation gap nobody wants to admit
The numbers tell a stark story. Research conducted by Analysys Mason observed that although 87% of communications service providers (CSPs) have started to implement AI in network operations, only 6% consider themselves to have achieved an advanced level of automation towards zero-touch operations. Meanwhile, Gartner’s Network Automation Market Guide 2025 report states that 67% of enterprise networking is still manual, resulting in slower change cycles, greater risk and increasing operational complexity.
There is a compelling business case to close that gap. Optical network automation can reduce the costs of management of the system lifecycle by up to 56% and operational costs by up to 81%. Provisioning times that used to run to ten days can be reduced to around 24 hours. As operators grapple with the complexity of 5G, network slicing, edge computing and AI-driven services, these are existential imperatives.
Yet Gartner cautions that by the end of 2028, more than 80% of full network automation efforts will have been abandoned, not because the underlying AI models failed at that point but because the organisations employing them weren’t able to iterate quickly enough to make them work. The hurdle is not technological, but more structural.
How legacy operating models create an automation ceiling
Most large telecoms networks were built around waterfall delivery covering long planning cycles, large project gates and sequential handoffs between network, IT and operations teams. For decades, this made sense. Networks were physical assets, changes were infrequent and the cost of errors was enormous.
That model is now actively obstructing the path to autonomous operations. McKinsey describes the consequences plainly, as basic network functions require long lead times, capacity cannot be scaled dynamically and processes remain complex and rigid. The firm is direct in its assessment, ensuring that end-to-end digitisation across the network lifecycle, as opposed to the tactical automation of isolated processes, is essential, yet waterfall-structured organisations are structurally ill-equipped to deliver it.
The pattern repeats across the industry. Technology is layered on top of old models of operating rather than transforming them in a fundamental way. Departments such as network, IT and operations are segregated into isolated teams. Legacy OSS/BSS environments constrain end-to-end workflow execution. Manual change management practices, built for slower network cycles, delay real-time automation. Vendor-specific systems complicate orchestration across domains. The result is partial automation that still depends on human intervention to stitch workflows together.
McKinsey reinforces this by noting that the barrier appears not to be experimenting with AI. Across most telcos, that is still at an early stage but already underway. The real challenge is putting it into meaningful use. Almost three-quarters of operators currently apply AI to improve existing processes rather than rethinking how they operate, limiting the productivity impact.
Bain & Company’s analysis, drawing on TM Forum’s Autonomous Network Level Evaluation Tool data, identifies network change management as difficult, with a mean autonomy level of just 1.9 out of 5. That is not a platform problem. It is a workflow and decision-rights problem.
Agile delivery as real prerequisite
For operators making genuine progress towards zero-touch networks, the transformation typically begins not with AI platforms but with agile operating models. Across industries, McKinsey found that organisations conducting agile transformations were three times more likely to become top-quartile performers than those that had not.
The real difference between truly agile telcos and those mired in the lower rungs of automation boils down to a few fundamental structural elements.
Siloed handoffs are replaced with cross-functional ownership. Deloitte’s 2024 Telecom Industry Outlook revealed that telcos in particular want less silo and more cross-functional structures to support increasing technological demands. Without that transition, automation efforts are ‘stuck at domain-level experiments that are not integrated across end-to-end workflows.’
Iterative delivery replaces big-bang programmes. McKinsey advises operators to focus first on quick wins that build steady momentum and eventual scale, basically a pragmatic alternative to the comprehensive transformation programmes that frequently stall under their own weight.
C-suite ownership drives outcomes. Where leadership commitment to automation is visible and funded, projects achieve scale. Where automation is delegated to a technology workstream, it stalls. McKinsey’s recent c-suite survey found that 64% of respondents are focused on scaling up AI to capture a 10–15% improvement in EBITDA, but ambition at the top must translate into structural change below it.
DevOps and NetOps integration creates the technical foundation. Gartner foresees that the percentage of first-time network configurations made through generative AI will rise to 25% by 2027, up from less than 3% in 2024, but just in those environments where NetDevOps methodologies, such as the use of CI/CD pipelines and infrastructure-as-code, are already embedded.
Scale of what is at stake
Commercial urgency behind this transformation is difficult to overstate. Appledore Research estimates total telecom operational expenditure at US$1.38 trillion, a figure that cannot be managed through incremental process tweaks. It demands a structural rethink. Appledore further estimates that over US$130 billion in entirely new incremental revenues awaits operators capable of delivering highly agile and automated orchestration.
By itself, the telco AI market is expected by IDC to grow from US$235 billion in 2023 to US$632 billion in 2028. But the operators that are best placed to capture that value are not at all the ones with the most advanced AI stacks. They are those with the organisational agility to deploy, test and iterate quickly. ABI Research forecasts that telcos using generative AI for network design, optimisation and testing can achieve productivity gains of 15–25%, rising to as much as 40% in customer care and BSS applications.
What agile telcos are actually doing differently
Operators making measurable progress share recognisable patterns. They pursue domain-level automation first, before attempting cross-domain integration. Swisscom concluded that operational autonomy at the domain level was an essential precondition before tackling network autonomy more broadly. They view automation as a product and not a project, developing continuous delivery platforms as opposed to one-off solutions. They embed AI into closed feedback loops, beginning with fault and performance management, where results are quantifiable, rather than releasing it as a standalone capability.
At present, Asian telcos lead on measurable autonomy with an average ANLET score of 3.56, followed by the Middle East, Africa, Europe and Latin America. The average industry-wide level of automation maturity increased by a whole level between 2024 and 2025, a development for the most part led by Asian-based operators.
Workforce implications run deep as well. McKinsey Global Institute research suggests that about 30% of work hours in telecoms could very well be automated by 2030, while 91% of telecom positions will need to be redesigned. Alongside new roles, such as AI workflow architects, network data engineers and autonomous operations specialists, are also emerging, along with automation systems.
The road ahead
The takeaway is simple. Zero-touch automation is a transformation challenge, not a technology one. It is a transformation problem rooted in operating models, delivery processes and organisational culture.
The most advanced operators are nearing 70 to 80% end-to-end automation, across digital processes, with long-term targets of 80 to 90% automated service fulfilment in sight. But reaching those numbers requires something that no vendor can sell: a fundamentally different way of working. Clearly, zero-touch automation is the destination, while agile culture is one route to get there.
For telcos still in planning stages, the data sends a clear message. Start with the organisational structure, not the vendor shortlist.
Anna Ribeiro