Qualcomm details its AI-native 6G vision which spans the air interface, through devices and the network, to the cloud
Qualcomm is positioning 6G as much more than the next generation of wireless. During Mobile World Congress in Barcelona, the company and a broad group of ecosystem partners laid out a milestone-driven path toward pre-commercial 6G validation in 2028 and initial commercialization starting from 2029 onwards. The architectural premise is that 6G will be AI-native across three domains: devices, networks and cloud infrastructure. Qualcomm describes that future system as combining connectivity, wide-area sensing and high-performance compute into a new platform for services and business models.
That framing is key because AI-native does not simply mean bringing 6G to market as an AI-integrated network. In a discussion with RCR Wireless News, Qualcomm Global Head of 6G Research Tingfang Ji described a broader design philosophy spanning applications, devices, radio technologies, the network and a nascent distributed compute fabric needed to support inference at scale. “There are many facets of AI-native 6G,” he explained.
On the device side, Ji sees the smartphone evolving into something much more capable and contextual. “The phone itself is transforming,” he said, pointing to agentic devices that can carry out tasks on a user’s behalf. But he also stressed that the phone is only one part of a much wider device landscape. AI-enabled AR glasses, robotics and connected cars all fit into the same architectural shift in which AI becomes ambient and embodied. “All these agentic functions will come to devices,” Ji said. In that sense, 6G is being shaped for faster connectivity and a world filled with intelligent endpoints that perceive, assist and act.
The network side is equally important. Ji broke AI-native 6G into two broad classes. One is AI for the RAN, where agentic algorithms support functions like self-diagnosis, fault detection and self-healing. The other is AI in the air interface itself, including areas like coding, modulation, waveform, channel state feedback, beam management, positioning, and sensing.
Efficiency is a central theme running through that vision. Ji framed 6G efficiency in multiple dimensions, including spectrum efficiency, cost efficiency and coverage efficiency. On coverage, he highlighted Giga-MIMO as a way to extend higher-frequency coverage without forcing operators into blanket densification. “As frequency goes up, the wavelength goes shorter. So in the same physical space you can pack in more antennas…You could actually beamform better and better, so you could concentrate your energy toward a direction so you don’t have to densify.” From an operator point of view, this illustrated the interdependence between spectrum efficiency, cost efficiency and coverage efficiency.
Ultimately, the AI-native idea lands in the cloud and edge as much as it does in the device or RAN. The Qualcomm-led coalition sees 6G as reliant on virtualized and cloud RAN, energy efficient compute and both edge and centralized data centers. Ji’s explanation fills in the operator role within that system. “Devices cannot always be doing everything because they’re power limited,” he said. That creates a clear opening for inference offload and distributed intelligence. “The operator has a role to play. They’re not only delivering bits, they’re delivering compute and intelligence.”
