The global digital economy is increasingly shaped by the convergence of advanced mobile communications and artificial intelligence, according to a paper published by GSMA and GTI Telecom.
As 5G scales worldwide, mobile networks will expand coverage and improve service quality while AI shifts from centralized cloud systems to on-device and edge deployments. Pervasive mobile connectivity broadens access to AI services, and AI in turn reshapes network architecture and operations.
The report defines the ultimate outcome as “Mobile AI”—a collaborative device–edge–network–cloud system that pairs network reliability and low latency with AI algorithms capable of perception and decision-making. The architecture is described as three vertical layers (foundation, execution, and application) combined with four horizontal domains: AI for Networks, Networks for AI, Mobile AI agents/terminals, and Mobile AI applications.
The whitepaper emphasizes that Mobile AI will require global cooperation and shared standards. As 5G-Advanced and 6G mature, mobile networks and AI are expected to underpin large-scale intelligent services across industries.
The report in detail
The report’s central claim is that Mobile AI will arise from continuous interaction among devices, networks, edge computing, and cloud platforms, with mobile infrastructure both carrying traffic and supporting AI workloads. Mobile traffic driven by AI services is projected to grow rapidly: the paper forecasts AI-related network traffic increasing at a compound annual growth rate (CAGR) of over 70% over the next decade. By around 2031, AI traffic could surpass traditional application traffic on global networks.
Demand for edge AI processing will accelerate because edge inference depends on device-to-network exchanges, which place new requirements on uplink capacity and push for near-zero latency. These changes will force operators to rethink capacity planning, network design, and operational models.
The Mobile AI era
The report proposes a “device–edge–network–cloud” architecture. Devices handle local sensing and immediate inference; edge infrastructure performs low-latency computation and aggregation; clouds provide model training, long-term reasoning, and large-scale coordination. Telecommunications networks connect these layers and manage traffic, service quality, and security. The authors also note that networks themselves will need AI-driven optimization to meet performance and efficiency targets.
The architecture’s three layers are:
- a foundation layer that delivers connectivity, computing resources, and data infrastructure,
- an execution layer that packages those resources into deployable services,
- an application layer that delivers sector-specific solutions.
Its four functional dimensions are:
- AI applied to network planning and operations,
- networks engineered to support AI workloads,
- AI-capable devices and agents at the edge,
- application ecosystems built on these capabilities.
A significant portion of the report focuses on AI operating within networks. AI can improve network planning and operational optimization by using real-time data to adapt capacity and configuration. For operations and maintenance, AI systems can detect anomalies, predict faults, and coordinate responses—moving networks toward more autonomous management.
Networks supporting AI applications
Intelligent devices and agents will create new traffic patterns and higher volumes, along with diverse service requirements. Some use cases—such as robotics or remote control—demand ultra-low latency and deterministic performance; others—like video analytics and multi-sensor fusion—generate large uplink volumes. Networks must move beyond traditional best-effort connectivity to offer flexible, differentiated service models that match application needs.
To support these new models, the report highlights three key adaptations operators should pursue:
- enhanced uplink capacity to accommodate data-rich streams,
- differentiated quality-of-service (QoS) levels for latency-sensitive and high-throughput use cases,
- closer coordination among device manufacturers, telecom operators, and software developers to ensure interoperability and efficient deployment.
Device and edge mobile
Mobile AI devices and agents are expected to drive much of the demand over the next decade. Smartphones, wearables, robots, and industrial terminals will evolve from passive endpoints into intelligent systems capable of local reasoning and task execution. Today’s hybrid model—local models handling immediate tasks while the edge or cloud performs heavier reasoning—will persist, but connectivity will increasingly be treated as part of the overall IT topology rather than a standalone service.
Mobile AI applies across many industries, including industrial automation, smart manufacturing, connected vehicles, urban management, healthcare monitoring, and energy systems. In these domains, AI functions must operate close to physical assets and rely on networks for connection to remote compute resources and data stores. The next-generation device–edge–network–cloud architecture is intended to deliver that capability.
Challenges and paths forward
Infrastructure limits are a major challenge: AI services demand higher uplink capacity, lower latency, and strong reliability—capabilities current network designs may not provide by default. Spectrum choices, including mid-band and millimeter wave allocations, will influence whether networks can meet these stringent requirements. Operators should expect rising capital expenditures as they expand and optimize edge compute resources and densify networks.
Standards and interoperability are another concern. Fragmentation in device protocols and AI interfaces raises integration costs and risks limiting cross-vendor, cross-market services. The report calls for increased collaboration through international standards bodies and industry alliances to harmonize interfaces, reduce fragmentation, and lower deployment costs.
Commercial uncertainty remains significant. While Mobile AI could create new revenue streams—such as selling AI infrastructure services, offering sector-specific data products, and delivering AI-driven services to enterprise customers—these business models are still emerging. Realizing them will depend on technical cooperation across operators, AI firms, middleware and software vendors, and device manufacturers.
The ideal for Mobile AI
AI workloads, intelligent devices, and distributed computing will reshape network architecture and operator business models. The pace of change will depend on infrastructure investments, spectrum policy, industry collaboration, and the emergence of viable commercial services as AI-driven traffic grows. Operators that integrate connectivity, edge computing, and data-processing capabilities are positioned to capture a larger share of the value generated by AI-enabled digital services.
(Image source: “Telegraph Poles” by spratmackrel is licensed under CC BY-NC-SA 2.0.)
Interested in how IoT is transforming telecoms and connectivity? Attend the IoT Tech Expo events in Amsterdam, California, and London to explore how 5G, edge computing, and IoT innovations are shaping the future of networks and services. The events are part of TechEx and are co-located with other leading technology conferences.
Telecoms News is produced by TechForge Media. Explore other upcoming enterprise technology events and webinars through TechForge’s events listings.