Flexible networks in mobile AI era: From data pipelines to value platforms

With the integration of AI into mobile communications, the era of mobile AI is rapidly approaching. Innovative services, such as real-time multi-modal calling with AI assistants, autonomous robots with embodied AI, and real-time cloud rendering for AIGC based on spatial large models, are emerging as killer apps in the 5G-Advanced (5G-A) era. The focus of AI applications is shifting from human-to-human and human-to-machine interactions to interactions between intelligent agents. This is leading to an exponential increase in connected objects, and the focus of services is shifting from only guaranteeing downlink speeds for video playing to guaranteeing both uplink and downlink speeds for diverse video services.

These transitions have imposed demanding new conditions on mobile networks, requiring them to support ultra-high-definition (UHD) duplex video streams and guarantee a deterministic experience free from any lag, buffering, or waiting. Following the voice and data eras, we now find ourselves entering the mobile AI era, one where we need networks to deliver a guaranteed 20 Mbps uplink speed per user and 1 Gbps uplink capacity while ensuring a ubiquitous coverage to access AI-powered apps anytime anywhere, akin to near-human level interactive experiences.

Facing such changes, carriers must build an intelligent productivity network with the optimal total value of ownership (TVO) for mobile AI, while maintaining existing services. To this end, Huawei offers the GigaGreen and GigaBand solutions. They not only provide greater capacity, a seamless experience, and minimal energy consumption, but they also deliver high elasticity and multi-dimensional experience assurance.

Flexible networks supported by mobile AI

Global networks, especially those running on mid-to-low bands, often face serious congestion issues and the challenge of constantly rising OPEX (including electricity and tower rental expenses). Huawei's latest GigaGreen radio frequency (RF) solution introduces the world's first Massive MIMO (MM) on three low-bands, which are the world's first large-scale commercial application of 8T8R and MM on the three medium-bands.

Huawei GigaGreen RF is an energy-saving technology for mobile base stations that cuts power use while maintaining strong signal coverage. It uses AI to adjust signal strength in real time and combines 2G, 3G, 4G, and 5G frequencies into one efficient unit.

Compared to traditional solutions, GigaGreen increases capacity between four- and seven-fold, significantly improving user experience. Its innovative materials, components, and systems make equipment smaller and lighter. For example, the integrated EcoMatrix solution, covering the full Sub-3GHz band antenna, reduces wind resistance by 55% whilst power consumption by over 40%.

GigaGreen hardware can be combined with Huawei GigaBand, a next-generation broadband solution that delivers ultra-fast, reliable internet connections for homes, businesses, and even industrial apps, such as smart factories.

Mobile communications are evolving from 4G to 5G, and from NSA (4G at the core, with 5G used to boost speed) to SA (a dedicated 5G core). In this evolution, exclusive spectrum use by different radio access technologies (RATs) has compromised user experience. To address this challenge, GigaBand enables full spectrum pooling and dynamically allocates spectrum resources to different users in real-time, significantly improving the experience of users connected to 5G SA networks without degrading the experience of other users.

Here are some benefits of GigaBand:

Spectrum resource pooling: Unlike traditional spectrum management based on band segmentation, with GigaBand, we can allocate resources — where they are needed most to secure performance for users. Spectrum resource pooling integrates the resources of mid-to-low bands (such as 700MHz and 2.6GHz) and high bands (such as 3.5GHz) to form a unified resource pool capable of supporting dynamic spectrum allocation based on service needs. For example, low-latency services like AI video calls or gaming can invoke high-band resources in real time, while mid-to-low bands are prioritized for large-scale IoT connections, realizing "one pool for multi-services." GigaBand enables operators to guarantee these service levels for both, even at peak times.

Multi-RAT integrated scheduling: Through the use of soft bus technology that syncs multiple devices, GigaBand enables the collaborative management of 2G/3G/4G/5G multi-RAT networks. In June of 2025, its first commercial deployment happened in Hong Kong, Gigaband set a benchmark for network upgrade in the mobile AI era. Supported by spectrum pooling and multi-RAT integration technologies, the carrier smoothly deployed 5G SA networks on multiple Sub-3GHz bands, automatically optimized resource allocation between 4G and 5G.

This guaranteed an uninterrupted experience for 4G users' voice calls and mobile broadband services, and released more bandwidth for 5G users. Field test data showed that while maintaining a stable experience for 4G users, the average throughput of 5G users increased by 2.28 times, supporting smooth real-time interactive services such as AI New Calling and cloud gaming.

Operations optimization: GigaBand ensures an optimal user experience at any time by collaboratively optimizing the scheduling of time, spectrum, space, power, and RATs. For example, the network will schedule more low-band resources for gaming users who need lower latency. In the case of livestreaming users who need higher uplink speeds, the network will schedule more frequency, power, and beam resources to guarantee their uplink experience. This means GigaBand can guarantee on-demand, differentiated experiences to provide solid network support for HD live streaming, low-latency gaming, wide coverage scenarios, and multi-modal applications.

In the mobile AI era, networks are going beyond their traditional role as mobile data pipelines, and evolving into value platforms that underpin differentiated experiences. GigaBand represents a new direction for 5G-A network evolution. Rather than employing the simple overlay of technologies, the solution enables intent-driven networks with intelligent orchestration, establishing networks as the core digital infrastructure of the AI era. What were previously no more than bright outlooks for the mobile AI era, such as on-demand spectrum resource allocation (no different from water and electricity) and differentiated, targeted network assurance for individual AI applications, are now becoming a reality, thanks to these innovative technologies.