i卡會員

The adoption of AI has become mainstream across industries, and the communications sector responsible for critical cross-border data transmission is no exception. In fact, AI has been deeply embedded from the design stage. As evidenced at the 2026 Mobile World Congress (MWC) held in March, AI is no longer merely for value added applications. Rather, it has become the core logic underlying the architecture design for the communications industry. The shift from “AI + network” to an AI-native paradigm is not simply a technological upgrade, but a reshuffling of the industry value chain. 

First, at the level of Radio Access Networks (RAN), AI-RAN has emerged as a key development direction. AI-RAN refers to a revolutionary technology that directly embeds AI into the physical layer of telecom networks. In the past, RAN systems relied on manual configurations and fixed rules for resource allocations. Today, with embedded machine learning models, networks can predict traffic fluctuations in real time, automatically optimize spectrum utilization, and reduce energy consumption. The GPU-accelerated architecture jointly developed by Nokia and NVIDIA enables real-time computing of AI models on the edge, i.e., base stations. Qualcomm has showcased its autonomous network technologies capable of dynamically adjusting frequency band utilization and enhancing the flexibility of resource allocation during peak hours. This is significant because networks are no longer mere transmission channels, but intelligent nodes capable of judgement, orchestration, and optimization. 

As infrastructure becomes intelligent, business models naturally evolve. With 86 operators already signed up, the GSMA Open Gateway initiative allows third-party providers to directly access network capabilities via standardized APIs. Take financial transactions as an example, SIM card and device consistency can be verified in real time to reduce fraud risks. In smart manufacturing, the quality of low-latency private networks can be dynamically ensured. For connected vehicles, priority transmission channels can be provided. This signals a gradual transformation into platform enterprises for telecom operators as their revenue mix is shifting from simple traffic-based billing toward “capability monetization” and “API revenue sharing”.

Major telecom operators all over the world have also embarked on strategical initiatives. Deutsche Telekom has deployed multi-agent AI systems that enable networks in different regions to autonomously collaborate in handling congestion and failures. SK Telecom in South Korea has unveiled the 519-billion-parameter sovereign model A.X K1 and plans to build a 1GW-scale AI data center, aiming to secure computing power and data sovereignty. NTT in Japan is reducing energy consumption of transmission through its IOWN all-photonics network architecture and exploring quantum photonics technologies to address the massive energy demands of AI computing. These developments indicate that the telecom industry is deeply integrating with the cloud and computing power markets. Future competition will no longer be just about coverage and speed, but for the ability to integrate computing power and the depth of platform ecosystems.

This wave of transformation is also spilling over into the terminal device market. High-computing-power chips can support on-device inference of large models, enabling smartphones to perform real-time translation, image generation, and cross-app task integration even while offline. 

Agentic AI has become a new interactive interface, allowing users to integrate messaging, calendar, and search functions through natural language commands. That said, according to GSMA Intelligence surveys, AI only accounts for approx. 53% of device replacement motivations. Consumers still prioritize battery life, camera upgrades, and system stability. Whilst AI has become a technological focal point, it alone cannot drive the consumer market without high-frequency and stable use cases. 

The development trajectory shown at the 2026 MWC highlights a clear path of evolution: AI reshapes network architecture; new architecture results in changes of business models; and business models, in turn, influence terminal device design and user experience. The core value of the communications industry is shifting from “transmission efficiency” to “the output of intelligent capabilities,” with the center of value correspondingly moving upward.

For Taiwan, this transformation represents both a challenge and an opportunity for repositioning. In the past, Taiwan has played a key role in the supply chain of the global communications industry by offering strong capabilities spanning IC design, RF modules, and networking equipment manufacturing. However, in the AI-native communications era, the status quo as a supplier of hardware is likely to be gradually marginalized amid the trend for platforms and the drive for software. Going forward, the key lies in whether Taiwan can participate in AI-RAN and 6G standardization, integrate into telecom API ecosystems, and extend its chip advantages into edge computing and integrated solutions for computing power. 

At present, the development of the communications industry is no longer centered on innovation of a single product. Rather, it is about the integration of networks, computing power, and application ecosystems. The era of AI-native communications has already begun. The decisive factor is no longer the number of base stations, but the integration depth of intelligent capabilities and the degree of platformization.