Barcelona’s Fira Gran Via hosted an unusually restless MWC in early March, where the industry’s cheerleaders and alarmists gathered beneath the same glass roof. GSMA’s new director spelled out a blunt diagnosis: carriers must finish 5G standalone rollouts, prepare networks for an AI‑driven traffic surge, and harden every link against mounting data‑security risks. The warning provided a stark backdrop for a parade of Chinese booths that treated the trio of challenges as design briefs rather than crises.
Huawei’s stand was deliberate and doctrinal: reframe networks as intelligent systems rather than ever‑widening pipes. Its AI‑centric network demos — including an intent interface that automatically prioritises an 8K upload for a VIP at a packed concert — signalled a shift from selling bytes to selling predictable user experience. On the infrastructure front Huawei also showcased large SuperPoD clusters, claiming a “UnifiedBus” that ties thousands of accelerators into a single logical brain, a bid to create a non‑Nvidia compute pole for trillion‑parameter model training and large‑scale inference.
The show was as much about compute strategy as radio standards. Qualcomm framed the future as “AI‑native wireless,” with demonstrations that push AI into the physical layer and a tentative pre‑commercial timetable for 2028, while Huawei emphasised U6GHz products seen as a 5G‑Advanced to 6G bridge. Together the propositions mark a pivot in the industry: the next round of competition will be about embedded intelligence and interoperability, not only raw spectrum or peak headline speeds.
Consumer devices at the exhibition were noisy proofs of concept for a world where AI ships inside cheap, practical hardware. Honor’s Robot Phone — a camera gimbal married to an on‑device AI agent — and Alibaba Cloud’s Qwen Glasses, priced at RMB1,997, showed a dual strategy: premium robots and mass‑market AI wearables. Tecno/Infinix resurrected modular curiosity with magnetic add‑ons, while vivo, TCL and Xiaomi used optics, foldables and a concept car to signal capability and brand imagination rather than immediate mass production.
Robotics stole much of the visceral attention. Unitree staged physical humanoid sparring to showcase resilient control algorithms, while Agibot presented task‑focused machines designed to work in logistics, retail and manufacturing scenarios. ZTE’s plush iMoochi companion suggested an alternate tack: not every AI product must be hyper‑functional — some compete on emotional value. These demonstrations underline a Chinese playbook that mixes electronics, software and mechanical engineering to create tangible, monetisable products across price points.
The commercial angle was explicit: turn AI into something people can buy and use in daily life. Alibaba’s low price for Qwen Glasses and the show floor’s focus on hybrid “training + inference” compute stacks make clear that Chinese firms are pursuing vertical integration and aggressive pricing to grab adoption. That strategy plays to strengths in scale manufacturing, ecosystem bundling and close operator ties, but it also bumps against limits: advanced accelerators, high‑end chips and western software ecosystems remain chokepoints in places.
MWC 2026 did not produce a single technological victor, but it did sketch the battlefield for the next half‑decade. Chinese companies presented a coherent, full‑stack response to GSMA’s three “mountains”: smarter networks, alternate compute supply chains, and commoditised AI devices. For global carriers, regulators and rivals, the takeaway is clear — the debate is no longer whether AI will strain networks, but which vendors will supply the intelligence, infrastructure and devices that keep the digital economy running.