AI Initiatives at MWC Shanghai Drive Uplink Requirements

MWC Shanghai in June 2026 highlighted optimism for the future. This year’s event centred on the path to 6G, including the role of 5G-Advanced as a stepping stone, and AI everywhere. As at many recent events, there were numerous robots at the show, including predominantly humanoid models.

Despite clashing with DTW in Copenhagen, MWC Shanghai claimed 37,000 attendees from 143 countries. Chinese companies were present but their vision was global, aiming to take the leadership examples and innovation already present at Chinese operators into global markets.

The discussions about 6G and 5G-Advanced in Shanghai reflected this appetitive for growth. In 2029, when the first commercial 6G networks are likely to launch, FDM CCS Insight forecasts that 1.1 billion mobile phones will be shipped. More than 85% of smartphones will be AI-enabled by 2030. The growing use of AI and AI agents on smartphones will lead to greater network traffic and changes in data traffic patterns, which increase uplink requirements.

Other device categories will become more important: FDM CCS Insight projects that wearable device shipments will increase by 32% in 2030 compared with 2025 and will represent a further 250 million new mobile devices. New device types such as augmented reality display glasses will see growth, but AI smart glasses will grow even faster by a vast 263%. AI glasses will benefit from improvements in network uplink connectivity and more-responsive lower-latency networks.

Cellular-connected vehicles will also increase enormously, placing further load on networks. FDM CCS Insight forecasts that the installed base of connected cars will grow by 41% between 2025 and 2030 in North America, Europe and Australia. Connected buses will rise 88% and commercial trucks and vans will increase by 38% in the same period. Overall, 492 million connected vehicles will be in use.

Vehicles will use network connections for infotainment such as streaming music or video, telematics and increasingly for visual AI to help map the world through on-board cameras to improve autonomous capabilities. Like AI glasses, this visual element will require a consistently good uplink connection, which is one of the focus areas for 5G-Advanced and AI-native 6G.

6G Optimism Pervaded MWC Shanghai

At this year’s event I was struck by the degree of confidence of almost everyone in the timeline for early 6G deployments. In Barcelona, we saw 52 companies — coordinated by Qualcomm — commit to the goal of 6G deployment by 2029. In Shanghai, it felt like everyone was on board with the 6G vision and there were few dissenters.

There was even some discussion of moving more quickly with 6G, with pre-commercial roll-outs as early as 2028. The LA 2028 Summer Olympics are an obvious target for Western suppliers, but if so, I expect there’ll be matching initiatives in China at the same time.

New Uplink Network Requirements from Robots and AI

The makers of AI and robots expect that these devices will change the demands placed on networks. Physical AI, also known as visual AI, will be widely deployed on connected vehicles, robots and drones to help to understand, interpret and navigate the world. In this vision, photos and video streams will be uploaded into the cloud for AI processing and to tie world knowledge with the local environment. The media will also be used to create and maintain digital twins of the world. The stream of imagery will require consistent uplink performance everywhere, not only in city centres, otherwise digital twins will be incomplete.

At MWC Shanghai, humanoid robots were ubiquitous. Some of the demonstrations were somewhat frivolous, such as a robot football challenge that sought to tie into FIFA World Cup fever. However, humanoid robots from Honor, AgiBot and others were a common sight. Chinese companies continue to invest and innovate with embodied AI and humanoid robots. There’s clearly an expectation that many of these robots will be connected to 5G or 6G networks and will often offload physical AI processing, provided the cellular network performance is good enough.

AI agents will also alter network data usage. Now, most data traffic is triggered by humans, for example by choosing to watch a video stream, download an app, connect with businesses on WeChat or live stream to social media. With agents being deployed on smartphones, PCs, in enterprises and throughout networks, increasing amounts of data will be autonomously generated by agents connecting with services or other agents. This will cause a change to when and where data traffic flows, as well as to the total quantity of data.

Burgeoning Device Numbers Make Autonomous Networks More Important

With the arrival of AI, agents and new device types, the complexity of managing and operating networks will increase. There’ll be more customers of network operators, and some will be agentic. This means that operators will need increased flexibility and be able to resolve network problems more quickly.

Growing device numbers will also increase the complexity of operations and maintenance. Different service levels will need to be managed for each new type of connected device, usage or AI feature. Agentic approaches will help operators to better manage the network to match these needs.

Already, on 5G-Advanced networks we’re seeing deployments of AI agents to improve operations and maintenance. 6G’s AI-native design will take this to the next step. Operators that have already adopted 5G-Advanced and modern cloud architecture will be best placed to make a quick and simple transition to 6G.

Generating Revenue from AI Tokens

In telecom, there’s also now tremendous debate about the role of communication service providers in AI. Is there the potential for operators to sell tokens, rather than minutes or gigabytes of data?

Although I understand the draw of new business models, this looks to be a very challenging idea. It goes against the direction of the past 20 years of telecom for services to shift from metered to unmetered propositions.

Also, it’s unclear how operators could identify tokens travelling over their networks. If the AI model, or AI factory, is run by the operator it’ll prove easier. Then, the identification problem disappears and the model is similar to recent initiatives by the leading frontier AI companies. If open-weight AI models triumph, or at least compete on performance and cost with proprietary models, then telecom operators will be better placed to monetize tokens and run their own AI factories.

This open-weight AI model focus is certainly a trend now in China and one that Nvidia highlighted at GTC San Jose earlier this year. Alibaba’s Qwen is especially notable here. But if the AI frontier models remain proprietary, then it’ll be companies like DeepSeek, Tencent, Anthropic and Open AI that’ll be in the right position to monetize tokens. This is an area we’re tracking closely at FDM CCS Insight.

AI Connected the Key Announcements at MWC

Many of the major product launches tied into the event themes. For example, Huawei unveiled a GigaUplink solution that the company positioned as supporting mobile AI. There was a similar line when Huawei discussed the importance of the upper 6 GHz band as vital additional capacity to meet new AI-driven requirements. Additionally, on optical products, Huawei used its co-situated AI-ON Summit to market 10 Gbps optical products as central to growth in the AI era and aligning with China’s overall strategic plans.

On the concept of monetizing tokens, companies at MWC showcased new AI factory designs aimed at telecom operators. Huawei discussed what it called service-network-compute integration to enable new token-based business models. This means that with the enormous number of AI and agentic references are making it harder for each company to differentiate using AI. One approach should be for companies to provide more clear in-depth AI case studies.

Other Themes at MWC Shanghai

The GSMA discussed similar themes at MWC Shanghai as it did at Barcelona in March: the AI economy, 6G innovation, crossindustry digital transformation and intelligent networks with 5G-Advanced as well as non-terrestrial network (NTN) integration.

However, I was struck how speakers and exhibitors had little to say on sovereignty, likely because it isn’t a close fit with the technology-centric optimism of MWC Shanghai. Similarly, despite the hype and large capital investment going into satellite networks, NTN wasn’t a top trend. That doesn’t mean satellite networks aren’t important, but they ranked lower than AI, the path to 6G and other topics this time.

Coverage was also a missing theme. In China, the networks are so widely deployed and so high quality, there’s little potential to differentiate on the extent of a network. This reflects the success of Chinese operators, but it’s also true in many other markets throughout Asia, and sits in stark contrast to the network quality and coverage issues that plague many Western operators. It also explains the lesser emphasis on satellite — if a terrestrial network is present, satellite networks lack the capacity and indoor signal reach to compete.

However, coverage discussions must stop focusing solely on downlink performance. For AI devices, having good uplink coverage will be more important. The industry needs to adjust how it measures and reports coverage to assess the network’s real capability.

The refreshing technology optimism everywhere in Shanghai has been built on a strong network investment foundation. Although MWC Shanghai has a reputation as a China-focused show, that perception is misleading. Operators and suppliers in other regions should look to what’s happening here and evaluate how to apply the best innovations in their markets.