By Claus Hetting, WiFi NOW CEO & Chairman
Commercial Wi-Fi 8 products are unlikely to ship in volumes for a couple of years at least – but that’s not stopping major chipset vendors from introducing Wi-Fi 8 technology now. Following Broadcom’s Wi-Fi 8 announcement in October (and MediaTek sampling their Wi-Fi chips since October) Qualcomm now says it will announce a full portfolio of Wi-Fi 8 platforms at Mobile World Congress Barcelona in 2026. This time it’s all about delivering the sophisticated and reliable connectivity needed to feed agentic AI, the company says.
Agentic AI is on everyone’s mind these days – and the connectivity to support the accelerated demand for AI data will shortly be needed, in particular as AI-related traffic is expected to reach 15-20% of total Internet traffic by 2030. Add to this 17-21 connected devices per US household, 1.4 billion smart home devices shipped this year alone, and a five-fold growth in video streaming by 2034. The numbers are from Qualcomm’s latest blog post, which also reveals that Wi-Fi 8 chips from Qualcomm will be announced at MWC Barcelona 2026 this coming March 2-6.
This means that three leading chipset vendors will have announced Wi-Fi 8 SoCs exceedingly early – at least when compared to previous generational product cycles for Wi-Fi. Is Wi-Fi chipset competition increasing in ferocity or are companies simply gearing up early to address the expected perfect storm of increased connectivity demand and agentic AI everywhere? We’re guessing that both are true coupled with nobody wanting to be perceived to be behind even in the slightest when addressing the potential explosive onslaught of agentic AI in the home and elsewhere.
Qualcomm is already deeply invested into the AI space with their current Wi-Fi 7 platforms. The company’s first ‘AI at the edge’ Wi-Fi 7 platform stole the show at NetworkX in Paris in 2024 (we wrote about it here) while at this year’s NetworkX, Qualcomm added a 24 TOPS Dragonwing Network AI module to complement last year’s 40 TOPS version. The module will “unlock AI-driven network performance gains through features that identify apps and devices, provide real-time summaries of network issues and more,” Qualcomm said here. Full marks go to Qualcomm for focusing their AI solutions on delivering real, measurable QoE benefits to end users.
A second critical component of Qualcomm’s current AI-at-the-edge architecture is a set of AI-powered tools enabling connectivity performance optimisation. Qualcomm calls this the ‘Dragonwing Network AI Suite’ and it includes ‘Traffic Sense’ for classifying latency-sensitive traffic, ‘Anomaly Sense’ to identify performance issues, ‘App Sense’ for identifying apps and tuning app performance. This new AI suite can be viewed as the intelligence feeding Qualcomm’s Software Defined Wi-Fi architecture released in connection with the company’s first Wi-Fi 7 platforms in 2023 (read more here).
World’s first consumer-grade AI router from Xiaomi
In September of this year Qualcomm and Xiaomi launched what is arguably the first consumer-grade ‘AI router’ featuring Dragonwing Traffic Sense for application-aware optimisation. “The Xiaomi BE10000 Pro empowers users with five one-click AI acceleration scenarios: Gaming, streaming, video calls, live broadcasts, and online classes. With a single tap, users can tailor their network for optimal performance, leveraging Qualcomm Traffic Sense for real-time, intelligent traffic management,” Qualcomm says. Qualcomm and Xiaomi are among the finalists at this year’s WiFi NOW Awards. All the winners will be revealed at the traditional WiFi NOW Awards Gala this coming Thursday.
Qualcomm has also been instrumental in bringing the Ray-Ban Meta smart glasses to market, arguably one of the most enticing new gadgets of this year and certainly one of the most practically useful XR devices ever. The glasses feature Qualcomm’s Snapdragon AR1 Gen 1 platform including Wi-Fi 6 connectivity. Specifically in the case of wearables, it is critical that future AI processing is done away from the glasses themselves so as not to consume power and processing time. Instead most of the AI processing should be done at the edge, Qualcomm says.
/Claus.
