> **来源：[研报客](https://pc.yanbaoke.cn)** # Summary of AI Adoption in Telecom and the Road Ahead ## Core Content The telecom industry is undergoing a significant transformation driven by artificial intelligence (AI), moving from experimental phases to enterprise-wide deployment. AI is becoming a foundational enabler for modernising networks, enhancing customer experience, and improving operational efficiency. The report highlights the current state of AI adoption, key challenges, and the preferred strategies for successful implementation. ## Main Findings - **AI Adoption Phase**: Over half of the surveyed telecom operators are in the process of scaling AI across their organisations, with the majority in the pilot or exploration phase. - **Leading AI Adoption Areas**: - Networks & Operations - Customer Experience (CX) - **Agentic AI**: - 94% of operators are either piloting or exploring agentic AI use cases. - 40% have initiated pilots, while the rest are exploring potential applications. - **AI Investment Priorities**: - AI infrastructure (cloud, edge, on-premises) is the top investment focus. - Internal AI development is second, followed by vendor partnerships and talent acquisition. - **Regional Focus**: - Across all regions, networks and operations receive the highest share of AI investment. - CX and service assurance are the second and third most invested areas. ## Key Challenges - **Operational Readiness**: Has become the main constraint for AI adoption, surpassing technology access. - **Legacy Systems**: Continue to limit data integration and real-time decision-making. - **Data Quality**: A critical factor in scaling AI, with poor data quality and integration being major barriers. - **Skill Gaps**: A significant obstacle, with many operators lacking the internal expertise to implement AI effectively. - **Governance Complexity**: Adds to the challenges of AI deployment, especially in ensuring compliance and ethical use. - **Regulatory Risk**: Increasingly influences AI deployment decisions, with data sovereignty and compliance becoming key considerations. ## Preferred Implementation Models - **Hybrid AI Operating Models**: Most telecom operators prefer a hybrid model that combines centralised governance with decentralised execution. - **Cloud-First Strategy**: Dominates AI deployment, though interest in sovereign AI is growing. - **Outcome-Based Partnerships**: Operators are shifting towards partnerships that align with business outcomes rather than just technical delivery. - **Balanced Scorecard for AI Success**: Metrics such as cost savings, ROI, customer satisfaction, and model accuracy are used to evaluate AI success. ## Best Practices for AI Adoption - **Improve Data Quality**: Enhancing data availability, accessibility, and quality is critical for AI success. - **Build Internal AI Talent**: Operators should invest in centres of excellence and internal training to reduce dependency on external vendors. - **Anchor AI to Core Value Streams**: Focus AI investments on areas with measurable business impact, such as networks, operations, and CX. - **Adopt Hybrid Governance Models**: Combine centralised control over infrastructure and governance with local execution for agility and compliance. - **Define Clear ROI Expectations**: Success metrics should be defined upfront, including financial, operational, customer, and technical KPIs. - **Prepare for Regulatory and Sovereignty Requirements**: Proactive planning for data sovereignty and regulatory compliance is essential. ## Conclusion AI is no longer an optional capability in the telecom industry but a foundational enabler for future competitiveness. While investment intent is strong, the pace of adoption is influenced by legacy systems, data quality, skills shortages, and governance complexity. Operators are rethinking their strategies to balance innovation with control, and the focus is shifting towards hybrid models, outcome-driven partnerships, and people-centric approaches. The next phase of AI adoption will be defined by execution discipline and the ability to integrate AI into core operations effectively.