Beyond robot races: Inside China Unicom and Huawei’s innovations for the mobile AI era

Details: Staff Writer; Category: Operators; 06 July 2026; 383 views

One of most viral videos of 2026 to date captured a humanoid robot named Flash running in the Beijing E-Town Half-Marathon and Humanoid Robot Half-Marathon in April, and setting a world speed record in the process. What casual spectators may not have known is that the event was also a preview of the coming era of mobile AI – in this case, powered by China Unicom Beijing and Huawei Technologies.

In the simplest terms, the “mobile AI” era is just that – mobile networks harnessing the power of AI. But the actual execution is incredibly complex and requires massive amounts of innovation. The upcoming wave of intelligent services and intelligent devices powered by AI will place unprecedented demand on mobile networks that operators must be ready to meet.

Major network changes

According to Li Jie, President of Huawei Wireless TDD Product Line, the new network requirements presented by AI devices and services are resulting in three major changes, starting with the underlying logic of network traffic.

“In the past, OTT services aimed to ‘see’ the world. Today, AI services aim to ‘understand’ the world. This means AI services need stronger sensing capabilities that must be supported by networks, which raises whole different demands on latency as well,” Li says. “The traffic will no longer be just downlink oriented – both uplink and downlink capabilities are equally important.

A second major change is that network management evaluation standards will need to shift from being KPI-oriented to experience oriented.

“This is a much more complex evaluation standard, because it requires us to provide stronger network assurance capability and better scheduling algorithms that can reflect the actual user experience,” Li explains.

The mobile AI era will also require network construction standards to change to ensure mobile networks are aligned user needs. Huawei has proposed a “Three A” standard for network construction, in which the “A”s stand for awareness, analytics and action.

In practical terms, this means networks should be able to provide QoS assurance by being more capable in sensing the user experience more precisely, having a deeper understanding for user models, and reacting faster to AI service needs.

“More specifically, we will conduct user profiling to understand where a user is, what kinds of services they are accessing, and what kind of uplink, downlink, and latency requirements these services are demanding of our networks,” Li says.

Unicom’s mobile AI journey so far

China Unicom Beijing’s experience with the E-Town Half-Marathon confirmed the necessity of more robust uplinks in 5G-A to support AI-powered devices and services. The robots themselves required approximately 10-Mbps uplinks each for video uploads, environment sensing, gait control, and autonomous navigation. This meant Unicom had to reserve dedicated uplink slices to ensure a positioning accuracy of up to sub-decimetre level and an average end-to-end latency of less than 30 ms throughout the racing course.

Yang Lifan, deputy general manager of China Unicom Beijing, highlights this uplink capability as one of several major innovations the operator has deployed in the past year since it first announced its plans at MWC Shanghai 2025 to get ready for the mobile AI era.

“We are still at the very last phase of fine-tuning the performance of the network to make sure that the customers can get the very best result, but one thing I can share with you is that its scale will definitely be the largest in the world,” Yang says.

China Unicom Beijing has also made progress on deploying AI across its operations in the past year, with AI agents in its base stations conducting performance optimization automatically on a weekly basis (as opposed to a monthly basis, pre-AI). Meanwhile, on the O&M side, nearly half of Unicom Beijing’s 60 work items are executed automatically by AI.

“Our estimation is that by the end of this year 80% of the O&M work items can be handled by AI automatically,” Yang says.

Yang adds that Unicom’s incubator centre has spent the past year researching ways to evolve its network to meet the needs of the mobile AI era, such as leveraging the U6GHz band to add more spectrum capacity to its 5G-A network, and integrating satellite networks with terrestrial networks via the 3GPP’s Non-Terrestrial Network (NTN) standards.

Cross industry partnerships

The latter – like Unicom’s experience with the robot marathon – highlights both the importance and value of cross-industry partnerships in the mobile AI era. A recurring theme during this year’s MWC Shanghai event was the role of integrated satellite networks in the mobile AI paradigm to extend coverage everywhere.

“We need to collaborate with the satellite coverage providers, because in the future, the network will become an integrated network that includes space, air and terrestrial connectivity,” Yang said.

Another key area for cross-industry partnerships for Unicom is the low-altitude economy. In fact, Unicom has already deployed a system for low-altitude integrated sensing and communications around the Badaling section of the Great Wall, Yang says.

Yang notes that the Unicom’s understanding of the demands from robots and embodied AI has improved in the past year, as illustrated by its experience with the E-Town Robot Marathon.

“If we want to move forward, we still need to have more discussions, but we think we have a better understanding of the network demands for both low-altitude use cases and robots now,” he said.

More innovations for the mobile AI era

Huawei’s Li Jie said that the company is keen to continue to walk with Unicom and other operators on their mobile AI journey, and has been developing several innovations for that purpose.

For example, Huawei has been tailoring its wireless solutions to support the U6GHz band, which could add from 200 MHz to 400 MHz of extra capacity per carrier to give operators more radio capacity and build more efficient networks.

A related innovation is improving the coverage capabilities of U6GHz and higher frequency bands, which is typically 10 dB worse than its mid-band and low-band counterparts, Li says. “If we want to really maximize the value of the high band, we must find the ways to eliminate this coverage gap.”

To that end, Huawei introduced ELAA (extremely large antenna) technology and multichannel technology, which can help telcos maximize both coverage and their spectral efficiency at the lowest cost, Li says.

A third innovation for the mobile AI era from Huawei is Cell-Free technology, which aims to make the scheduling capability between multiple bands and multiple cells more collaborative with the goal of making the user experience consistent everywhere in the network – effectively making handoffs undetectable from the user’s point of view.

“Cell-Free means that from a network management point of view, we see the whole network as one cell, so then the user experience in the whole network is like they’re in one cell,” Li explains.