By the end of 2025, a clear inflection point has emerged: artificial intelligence is no longer just an application layer, but an infrastructure. This shift is redefining the telecom value chain. Networks no longer simply transport data—they orchestrate compute workloads. In this context, China’s ecosystem acts as an accelerated indicator of this new paradigm.
1. The global AI paradigm shift
AI is becoming a core workload for networks, introducing new constraints: latency, energy availability, and proximity of compute. The centralized cloud model is showing its limits as distributed inference scales. The challenge is no longer just to connect, but to process, prioritize, and monetize flows of intent. Infrastructure becomes intelligent by design.
2. China’s ecosystem: constraint as a catalyst
Under technological and geopolitical constraints, China has accelerated vertical integration across its stack: semiconductors, cloud, networks, and AI. This pressure has acted as a catalyst. Local players have built sovereign architectures optimized for energy efficiency and large-scale execution. The result is a strong ability to deploy AI clusters rapidly, bring compute closer to use cases, and industrialize real-world applications. Constraint becomes a driver of systemic innovation.
3. Weak signals & CAPEX opportunities for telcos
Several signals are emerging: the rise of regional data centers, hybrid edge/cloud architectures, AI embedded into network layers (AI-RAN), and above all, the convergence of energy and telecom. CAPEX priorities are shifting. Investing in spectrum is no longer sufficient; compute, storage, and energy are becoming critical. Telcos have structural assets—distributed sites, real-time capabilities, customer relationships—but must activate them within a platform strategy.
4. Competitive landscape: European operator responses
European operators are moving, but in a fragmented way. Some are exploring network APIs, others investing in edge computing or hyperscaler partnerships. However, the approach remains largely defensive. The risk is clear: becoming connectivity providers within a value chain captured by others. Faced with integrated models (China) and hyperscale dominance (United States), Europe needs to define its position.
5. Strategic recommendations: toward the “compute operator”
The transformation is structural and requires rethinking the operator’s role. Three directions emerge: integrate compute as a native layer of the network by developing distributed inference capabilities; monetize network signals through APIs and value-added services; and invest in energy as a key differentiator at the core of AI infrastructure performance.
The question is no longer whether to follow this transformation, but where to capture value. Tomorrow’s operator will not simply move bits, but orchestrate compute at the intersection of networks, data, and energy.
This work is based on continuous human monitoring, field insights, and one-to-one interviews with ecosystem players. It is designed for telecom, innovation, and infrastructure decision-makers seeking to anticipate the next inflection points, with analytical frameworks, weak signals, and actionable directions.
The document is available in French and can be provided in English upon request.