As AI moves deeper into telecoms, the conversation is shifting from what operators can build to how they can scale it. The real constraint is no longer imagination, but the foundations beneath it: the data, compute, and network intelligence that make large-scale AI possible.
Every AI use case, from customer engagement to new enterprise services, relies on this hidden layer. It is the industry’s engine room: the systems that train models, move data, automate decision-making, and enable real-time intelligence at the network edge. Its strength determines the pace of innovation, the reliability of outcomes, and ultimately how far a telco can progress along its AI maturity journey.
In this next part of our AI Transformation series, we look at this foundational layer, how operators are modernising their data environments, redesigning their infrastructure, and forming new partnerships to power the products and services explored in our previous blog. As the industry moves beyond experimentation, it is this intelligent core that will decide who can turn AI into sustainable, scalable impact.
Data as the Foundation of Intelligence
Every AI system, no matter how advanced, begins with data. For telcos, this data is exceptionally rich: real-time network telemetry, location information, service usage patterns, and millions of customer touchpoints. What sets operators apart is not the volume, but the uniqueness of this data. The real value appears when data is properly organised, well governed, and able to move across different parts of the business.
Early in their AI maturity journey, many operators focus on cleaning, consolidating, and governing fragmented data sources. As they advance, data evolves into an enterprise-wide asset: integrated, accessible, and usable for cross-domain AI applications.
 Orange has done this through its Flux Vision platform, which uses anonymised and aggregated mobility data to support city planning, transport management, and retail analytics.
Intelligent Infrastructure: From Cloud to Edge to GPU
If data is the raw material of AI, infrastructure is the power source behind it. Training models, running inference, and enabling real-time decisioning requires an entirely new class of architecture: cloud-native, distributed, and built for high-performance compute.
This shift is visible in the recent Nokia–NVIDIA partnership, which integrates NVIDIA’s AI-optimised GPU capabilities into Nokia’s Cloud RAN. This allows AI workloads to run closer to the network, improving performance and reducing energy consumption.
STC is pursuing a similar direction through large-scale investments in high-performance data centres across Saudi Arabia, designed to support AI workloads and cloud-native applications. These facilities are part of a broader national push to expand sovereign digital infrastructure, enabling future AI services in sectors such as healthcare, logistics, and smart cities.
Advancing along the maturity curve often involves moving from traditional on-premise systems to hybrid AI architectures: blending hyperscaler flexibility with operator-controlled edge compute. This hybrid approach is key to supporting emerging AI models, including generative and agentic systems that need low latency and real-time data access.
In short, infrastructure is no longer a utility. It is becoming an intelligent network layer, where compute, storage, and connectivity converge to support AI-driven value creation.
Monetising the Foundation: Where Data and Compute Create Ecosystem Value
As operators progress in their AI maturity journey, the same data and infrastructure used to power internal transformation can be opened up as market-facing capabilities.
In Japan, KDDI shows how AI-ready infrastructure can become a commercial asset. Working with AWS, the operator offers multi-access edge computing that lets enterprises run low-latency AI applications in areas like logistics, robotics, and smart manufacturing, turning network and compute capabilities into new revenue streams. Meanwhile, KT continues to expand its GPU-as-a-Service model, lowering the barrier for businesses to experiment with and deploy high-performance AI.
This is also the layer where agentic AI begins to emerge. Systems that analyse data, make decisions, and act autonomously. Whether coordinating logistics flows, adjusting energy distribution, or powering mobility services, agentic AI requires real-time access to both network intelligence and distributed compute. Telcos that build this foundation will be pivotal in enabling these next-generation use cases.
By monetising data, APIs, and compute, operators move from connectivity providers to AI platform players, capturing value across multiple industries.
Partnerships that enable scale
Scaling AI-ready data and infrastructure is capital-intensive and technically complex; no operator can do it alone. The next phase of transformation is being shaped through deep partnerships between telcos, vendors, hyperscalers, and AI specialists.
The Nokia–NVIDIA collaboration demonstrates how network expertise and AI innovation can come together to create AI-native infrastructure. Similarly, AT&T’s integration with Microsoft’s Azure Operator Nexus supports hybrid cloud environments optimised for AI operations. Telefónica’s work with IBM on open hybrid cloud architectures further illustrates how strategic alliances accelerate operator readiness.
Progress along the maturity curve often follows the depth of these partnerships. Early-stage operators rely more heavily on vendors and cloud providers; more mature operators co-develop architectures, share R&D, and shape emerging standards. Partnership has become more than an operational enabler; it is now part of the infrastructure itself.
The Path ForwardÂ
Data and infrastructure are becoming the decisive layers in AI transformation. They shape how quickly operators can deploy new capabilities, how reliably they can scale them, and how far they can progress along the AI maturity journey. As operators modernise their data environments and build AI-ready networks, they unlock opportunities that were previously out of reach, from real-time automation to new B2B services and ecosystem value.
Looking ahead, the operators that strengthen these foundations and invest in interoperable and energy-efficient architectures will be best placed to capture the benefits of AI.Â