AI Boom Forces Long-Delayed Wi-Fi Overhaul
▼ Summary
– Cisco’s report finds a gap between companies’ AI ambitions and their reliance on outdated Wi-Fi 5 (802.11ac), which is still the most used standard despite being released in 2013.
– AI-driven cyberattacks targeting wireless networks are increasing, while AI workloads in enterprises are expected to grow from 28% of respondents today to over 75% by 2027.
– Wi-Fi 6E’s use of the 6-GHz band makes it better for high-bandwidth AI tasks, and organizations using it have nearly double the rate of AI applications (45%) compared to non-adopters (26%).
– Wi-Fi 7, introduced in 2024, improves efficiency with Multi-Link Operation and wider channels, and is expected to dominate new enterprise deployments over the next two years.
– The upcoming Wi-Fi 8 standard will add AI inference engines to chips for local processing and multi-AP coordination, addressing the “wireless AI paradox” of AI driving both demand and cyber threats.
The rapid expansion of artificial intelligence within enterprise environments is exposing the limitations of outdated wireless networks, forcing a long-overdue upgrade cycle for Wi-Fi infrastructure. According to Cisco’s latest State of Wireless Report, companies are caught in a widening gap between their ambitious AI plans and the aging Wi-Fi standards they still rely on.
The report highlights a stark disconnect. While 28 percent of organizations already run AI workloads,a figure projected to surge past 75 percent by 2027,most remain tethered to Wi-Fi 5 (802.11ac). First released in 2013, this standard still dominates at 43 percent of respondents. Yet fewer than one in five companies have upgraded to any Wi-Fi standard introduced this decade, such as Wi-Fi 6E (2021) or Wi-Fi 7 (2024). Wi-Fi 5 tops out at 3.5 gigabits per second, a speed that struggles with high-bandwidth, low-latency demands in device-dense environments,a critical shortfall for streaming and AI applications.
“As network demands increase, Wi-Fi 5 will become more costly to operate,” said Matthew MacPherson, enterprise wireless CTO at Cisco. “Administrators will spend more time reacting and troubleshooting, less time applying tools for better experience, productivity, and security.”
The evolution of Wi-Fi standards has gradually addressed these challenges. Wi-Fi 6 introduced orthogonal frequency division multiple access (OFDMA), splitting signals into lanes to serve multiple devices simultaneously. However, MacPherson noted that Wi-Fi 6 was not originally designed for the complex, high-bandwidth traffic patterns AI now drives. Wi-Fi 6E unlocked the 6-gigahertz band, adding dozens of data channels and boosting suitability for crowded, high-bandwidth tasks. Organizations already using 6 GHz report nearly double the rate of AI applications (45 percent) compared to nonadopters (26 percent).
“The 6-GHz band supplies the bandwidth required by AI-powered applications and correlates with improved scalability,” MacPherson added.
Wi-Fi 7, released in 2024, introduced Multi-Link Operation (MLO), enabling devices to use multiple bands simultaneously for improved efficiency and stability. Wider 320-megahertz channels and better 6-GHz utilization deliver lower latency and more consistent performance. Gaurav Jain, vice president of technology at the Wi-Fi Alliance, emphasized that “access to uncongested 6-GHz spectrum around the world is critical to unlocking Wi-Fi 7’s potential.” Cisco expects Wi-Fi 6 to become the new baseline, but Wi-Fi 7 will capture most new enterprise deployments over the next two years as the equipment ecosystem matures.
Looking ahead, Wi-Fi 8 promises to embed AI capabilities directly at the network edge. Coordinated by the Wi-Fi Alliance, the standard is not expected until late 2027 or 2028. Its features will include multi-AP coordination (MAPC), dynamically serving the most urgent traffic across access points. “Wi-Fi 8 extends connection stability to the network level, enabling seamless roaming without any hit to the client,” MacPherson said. “It will bring network-level fluidity and more local processing power.”
A major engineering shift accompanies Wi-Fi 8: silicon manufacturers are integrating AI inference engines into chips. Broadcom unveiled the first such chip in late 2025, ahead of formal certification. This allows tasks like anomaly detection, spectrum optimization, and traffic prioritization to run locally on the access point, eliminating round trips to the cloud.
The urgency is compounded by what Cisco calls the wireless AI paradox: the same AI transformation driving demand for better Wi-Fi is also generating AI-driven cyberattacks targeting wireless networks. Better infrastructure, the report concludes, is no longer just about bandwidth. It is about building networks smart enough to defend themselves.
(Source: Ieee.org)
