The way the telecoms sector is deploying AI is undergoing a meaningful shift. Generative AI, in particular, is now being evaluated not just for what it can do, but also for how and where it adds value. While the global market for generative AI in telecoms is expected to exceed £9.8 billion by 2034, Communication Service Providers (CSPs) are actively exploring targeted deployments across assurance, customer care, planning, and analytics. This indicates that growth will be driven not by hype, but by disciplined execution.
CSPs emphasis on areas where real-time pattern recognition and process acceleration translate into measurable impact also drives a strategic focus on identifying high-value use cases. This involves aligning architectures to ensure compliance and addressing performance and data quality from the start.
However, as AI becomes deeply embedded within telecom operations and takes on a larger role in shaping network management, operators must also confront the growing ethical and regulatory challenges accompanying this rapid adoption. As AI rapidly shifts from a point solution conversation to one around capability development, the operators who are able to move fastest will be those who combine automation, governance, and cross-domain scale within a secure and transparent framework.
Beyond efficiency: AI as a network performance lever
AI is already improving how operators manage their networks. Whether in multi-vendor environments or dense urban deployments, machine learning models are enabling more accurate traffic predictions, automated troubleshooting, and dynamic energy optimisation.
CSPs can efficiently utilise ready-to-use AI or GenAI applications for fixed and mobile networks. For example, operators are achieving up to 30% reductions in carbon footprint through AI-driven optimisation of both active network elements and passive infrastructure such as cooling systems—without compromising customer experience. However, for CSPs with their own analytics and AI teams, wanting to build their own AI applications, there are telco specific data and MLOps solutions available too.
One area with significant traction is predictive maintenance. AI can now integrate telemetry with fault history and external variables to identify failure signatures in advance. This allows field teams to intervene before service degradation occurs, improving time to resolution, reducing operational overheads, and importantly ensuring a superior customer experience.
Consumer expectations are arguably at an all-time high, with customers demanding fast, seamless, and personalised experiences across every touchpoint, which also creates a remarkable new avenue for AI deployment. As consumers are more willing to share personal data for better experiences and personalised services, operators now have access to a vast amount of customer data, including real-time insights into behaviour, and preferences and usage patterns with AI. All this data allows operators to segment their audiences more effectively and deliver targeted promotions, customised service offerings that drive deeper engagement and long-term customer satisfaction.
Operators are also applying Gen AI to speed up support and assurance workflows and is being used for example for root cause analysis and predictive maintenance. This helps reduce manual input and enables agents to focus on more complex cases. With security also a critical focus, AI helps detect fraud patterns, synthetic traffic, and outliers in billing systems by identifying deviations at speed and scale. The ability to run these models at the edge adds an extra layer of protection in latency-sensitive or decentralised environments.
Strategic execution at scale
Scaling AI across domains remains one of the sector’s most significant challenges. Fragmented data environments, inconsistent tooling and low organisational readiness often create friction beyond initial pilots.
Operators that are making the most progress have taken a deliberate approach by defining a limited number of use cases aligned with business metrics. They are assessing organisational maturity and focusing on interoperability as well as investing in cross-functional teams that combine network expertise with data science and compliance skills.
Architecture matters too. AI’s positioning in the stack – whether embedded in OSS/BSS systems, deployed at the edge or integrated via APIs – has implications for scalability, governance and cost. As a result, reusability and modular design are now among the most important considerations.
Responsible AI: compliance, transparency, and trust
With the EU’s AI Act setting requirements for transparency, explainability, risk classification, and data handling, AI governance is becoming a core component of telecoms operations. For telecoms, this intersects with existing obligations under GDPR and sector-specific regulations.
Operators are already working to meet these expectations. Privacy-preserving techniques such as federated learning, differential privacy, and synthetic data generation have already been introduced. Some are establishing internal review mechanisms to monitor for bias, especially in customer-facing AI use cases such as credit scoring or eligibility routing.
However, transparency remains a critical factor for long-term confidence in automated systems. Model explainability, version control, and audit trails are now being built into design requirements, particularly for high-impact domains like network management and service automation. Undoubtedly, AI cannot be scaled without governance. Telecoms operators are accountable for how AI is applied across their systems, even when using third-party models or external platforms. This makes internal control mechanisms essential.
Operators are now building AI governance frameworks that include model approval processes, performance thresholds and fallback policies. These frameworks ensure models operate reliably and meet legal and operational standards, even as environments evolve.
While the EU AI Act introduces formal obligations for documentation and monitoring, beyond compliance, these practices improve internal alignment, reduce duplication and build trust in AI-generated outputs. As a result, governance enables the ability to scale AI confidently across customer care, network operations and beyond.
The changing shape of telecoms talent
The growing adoption of AI is also changing team structures and workflows. As repetitive tasks are being reduced and operational teams are starting to work with AI-generated insights, this is raising new questions about skills, accountability and role evolution.
Operators are responding by investing in AI fluency across business and technical functions. This includes prompt engineering, model oversight and the ethical use of automation. Some are also exploring new roles around AI service management, overseeing model lifecycle, training inputs and integration into daily operations.
While some functions may see reductions, others will expand. As AI automates routine diagnostics or workflow triggers, there is a growing need for people who can supervise complex processes, interpret recommendations and manage exceptions. Therefore, the focus now is on redesigning how work gets done and ensuring that teams are equipped to manage that transition.
A path ahead
AI is already playing a material role in network performance, customer service and operational efficiency. However, its long-term value depends on how well it is integrated, managed and scaled.
Having supported many operators throughout their AI journeys, we’ve seen first-hand that responsible deployment is essential. It must be a top priority to align AI adoption with business objectives and ensure that every implementation meets expectations for transparency, security and performance.
Only with the right structure in place, AI can become a stable, scalable component of telecoms operations, deliver value and adapt as technologies and regulations evolve.
