Telcos are facing growing pressure in a saturated connectivity services market, write Sarah Woon and Matt Duke-Woolley from Beecham Research. In spite of holding a crucial role in communications across industries, core products are increasingly seen as commodities, leading to the expected continuing decline of average revenue per user (ARPU) over the next few years. Whilst global telcos’ market revenue will grow, it is set to be at a sluggish pace, inhibiting profit margins and sector development. Therefore, telcos must seek ways to reduce unnecessary costs and unlock new revenue streams.
Operations are also a challenge: with worldwide data traffic increasing 24% year-on-year, expected consumption will reach 1.8 petabytes by 2026, increasing the complexity of network routing, congestion and surveillance. Artificially inflated traffic by fraudulent adversaries further exacerbates this problem, although global enterprise losses to this issue are expected to reduce by 8% in 2025.
To combat these issues, telcos are increasingly turning to AI. With a recent survey finding that 65% of telcos have already implemented an AI strategy, and another stating that AI (excluding GenAI), automation and machine learning is anticipated to be the top priority investment area in 2025, telcos are already making strides in enhancing their operations and business models through more efficient, dynamic and intelligent networks.
But AI goes beyond optimisation – it opens the door to real transformation.
With it, telcos can shift from being pure connectivity providers to intelligent platform operators. In this way, telcos have the potential to redefine their role in the broader IT and OT ecosystem, and with it, generate new and innovative opportunities for revenue growth.
Enhancing network automation and operations
AI fundamentally alters how networks are operated and experienced. The intelligence aspect of the technology means it has data-driven capabilities that assist with decision-making processes, such as how to best optimise networks and predict maintenance requirements.
AI-enhanced network automation is particularly critical in achieving these benefits since it offers the computational power, scale and speed lacked by traditional methods.
Key network automation features include:
- Self-optimising networks (SON): AI automatically adjusts parameters such as antenna tilt, power levels and handover thresholds to improve coverage and minimise manual tuning.
- Predictive scaling: AI forecasts traffic patterns and provisions resources in advance of demand peaks, reducing congestion and enhancing efficiency.
- Intent-based networking (IBN): Operators define performance outcomes for specific applications and the network intelligently adjusts itself to meet those requirements.
- AI-managed 5G slicing: AI allocates and adjusts network slice resources in line with service-level needs supporting highly differentiated services from autonomous transport to IoT telemetry.
Through deployment of these network features, telcos can combat major operational challenges to do with data volumes, bandwidth management and energy consumption, resulting in lower associated costs. It also improves network performance and reliability, ensuring more resilient telco operations.
Reimagine telco architecture for increasingly agentic networks
With nearly 70% of telcos exploring the opportunities of edge computing, the cloud is no longer the only feasible option for AI. Thanks to increasing edge computing power, deployment of localised AI offers opportunities for greater speed and autonomy of operations, as well as the development of new time-critical use cases.
However, intelligence does not need to be siloed within the edge or the cloud. Instead, there is huge potential for redesigning telco architecture as an interconnected web of small and large language models (SLMs and LLMs) that autonomously interact with each other and the cloud, each optimised for different tasks. For instance, lightweight SLMs deployed in cell towers could handle immediate diagnostics and network adjustments, whilst LLMs in edge-based data centres could provide deeper analysis and pattern-based insights. This frees up capacity in the cloud to perform tasks such as aggregating multi-site data, running complex analytics and training further AI models.
A redesign of infrastructure empowers telcos to shift from manual configuration and reactive operations towards an agentic, proactive communications ecosystem.
Expand telco offerings with AIaaS
AI-as-a-Service (AIaaS) is the provision of AI tools and capabilities as on-demand services, usually via subscription models. This enables organisations to bring AI into their products or workflows at speed, and without the upfront cost or complexity of developing their own infrastructure.
AIaaS may include machine learning models, natural language processing or computer vision, embedded within edge or cloud infrastructure, and accessible through application programme interfaces (APIs) or simple platforms. While some offerings may be generalised, many can be delivered as vertical-specific solutions, like those for smart factories, surveillance or connected vehicles.
With telcos already developing advanced AI capabilities to serve their own networks, they are well-positioned to become trusted AIaaS providers for their customers.
Generate value for enterprise users
In many cases, AIaaS can provide clear, measurable value to enterprise operations. For instance, AI-enabled predictive maintenance supports the forecasting of equipment failure. This reduces the risk of downtime and removes reliance on scheduled inspections – saving money, time and resources.
In other cases, it supports organisations in quickly developing and implementing new applications. For example, computer vision embedded in CCTV enables intelligent traffic management and real-time identification of crimes. Similarly, in manufacturing, AI running on MEC platforms allows for autonomous quality inspection and robotics control. Accelerated time-to-market is frequently a key factor for project success, impacting profit margins and ability to scale.
As a host of AIaaS, telcos deliver customers far more than just connectivity services – they create opportunities for enterprises to innovate and improve business value. This would shift the market positioning of telcos from commodity providers to business and intelligence enablers.
Unlock service differentiation and new revenue opportunities
With optimising customer experience being the top telco AI use case today, service differentiation remains a priority. Customising user experiences by sector, location or business size assists telcos in improving customer engagement – a major factor in reducing churn and increasing lifetime value per connection.
AI-enhanced network-as-a-service is a key enabler of this, since it facilitates dynamic network configurations and traffic prioritisation, tailored to the customer’s needs. It can also provide greater performance guarantees through monitoring and adjusting service parameters in real-time.
In addition, AI can transform network and user data into actionable insights for monetisation. For example, AI analysis of behavioural and contextual signals can be sold to enterprises to help them understand how to best meet their customer needs. Similarly, partnerships with insurers, regulators or government bodies will enable the generation of data-informed and usage-based policies.
Other data monetisation applications include mobility analytics for transport and logistics, footfall tracking in retail, and predictive maintenance services for industrial clients.
The revenue opportunities of AI-enhanced APIs
Network APIs are proving an increasingly important value tool to telcos, with the global market value set to soar to US$34 billion in 2030, representing a 34% CAGR between 2023 and 2030.
A major driver of this growth is the shift towards Open APIs, which generate revenue by letting third-party developers access, interact with and control aspects of telecoms network features. Frameworks like the TM Forum’s Open Digital Architecture (ODA) and the GSMA Open Gateway Initiative support this.
Telcos are increasingly enriching these APIs with AI-derived intelligence, such as movement forecasts, congestion detection and predicted user experience metrics, to make them more valuable to third parties. For these features, a higher fee can be charged.
The same goes for other telco-hosted platforms – the more intelligent and feature-rich the platform, the greater the opportunity for revenue generation.
Smart cities: Use cases for the telco AI opportunity
Smart cities represent a key vertical for the deployment of AI. As highly complex ecosystems composed of interconnected technologies, diverse infrastructure and public and private stakeholders, they demand intelligent coordination, real-time decision-making and resilient connectivity.
At the foundation layer of the solution is the telco’s own infrastructure. The use of AI internally for automation and efficacy of network operations enables efficient traffic routing, proactive fault detection, real-time predictive maintenance, and smart energy optimisation. This ensures a seamless and secure connectivity – critical to the digital infrastructure of smart cities.
Building on this, telcos can offer AI-as-a-Service to ecosystem partners, including city councils, public transport operators, utilities and retailers. By hosting pre-trained models and delivering scalable AI functionality, telcos can support partners in deploying intelligent, data-driven applications across the urban environment. Though many are cloud-hosted solutions, these capabilities are increasingly delivered at the edge to support low-latency requirements and preserve data locality for privacy-sensitive operations.
At the top of the stack, telcos can differentiate their services and generate new revenue by exposing their network intelligence through APIs and service-level tools. These programmable services support differentiated experiences for sectors such as retail, logistics and transport, positioning the telco as a strategic partner in delivering smart city outcomes.
