Broadcom announced the industry’s first Wi-Fi 8 access point (AP) and switch solution designed with a unified architecture for enterprise networks supporting AI workloads. The solution builds on Broadcom’s Wi-Fi 8 radios introduced in October 2025 and includes two new components:
The BCM49438 accelerated processing unit (APU) chip, which powers the enterprise-grade Wi-Fi 8 AP platform and combines high-performance computing, networking, and AI acceleration. It works with the latest Wi-Fi 8 radio chips: BCM43840, BCM43844, and BCM43820. The Trident X3+ BCM56390 Ethernet switch platform, which supports up to 48 multi-gigabit ports and integrates Broadcom’s multi-gigabit PHYs (BCM84918, BCM54908, BCM54908E) and PoE power sourcing equipment (PSE) chips.
Together, the platforms form a unified wired and wireless architecture intended to deliver performance, efficiency, and security for Wi-Fi 8 deployments in enterprise environments.
Broadcom highlighted three industry-first features in the solution:
End-to-end MACsec encryption implemented across all switch ports and access points.
Advanced telemetry and analytics engines covering the full portfolio from Wi-Fi 8 APs to campus switching infrastructure.
Integrated Edge AI/ML engine on the Wi-Fi 8 AP for model acceleration without requiring a separate NPU or coprocessor.
The company stated that Wi-Fi 8 will impose greater demands on network infrastructure due to higher bandwidth needs, ultra-low latency requirements for real-time AI applications, and increased telemetry data from connected devices. The transition to Wi-Fi 8 coincides with enterprise trends including hybrid work models increasing uplink and downlink bandwidth usage, rising adversarial AI attacks requiring multi-layered security, AI factories needing high reliability and low latency, and growing focus on operating expense (OpEx) for network upgrades.
Additional capabilities of the unified architecture include Wireless Time-Sensitive Networking (wTSN) using 1588 Precision Time Protocol (PTP) for deterministic, ultra-low latency communication, and enhanced real-time network telemetry to support AI-powered network operations (AIOps).
The solution is currently sampling to early access customers and partners. Interested parties should contact their local Broadcom sales representative for samples and pricing.
Statements from partners include:
Kumar Srikantan, Vice President and General Manager, Campus, Arista Networks, noted that Arista’s Cognitive Campus platforms will leverage the new platform’s native encryption on all ports and high-bandwidth interfaces for Wi-Fi 8 access points, combined with Arista’s EOS software.
Dan DeBacker, Chief Product Officer, Infrastructure at Extreme Networks, said the platform enables AI-driven networking at the edge with built-in intelligence, automation, and security.
Sanjoy Dey, VP of Wireless Product Management, HPE, stated that HPE will use the Wi-Fi 8 platform for its next-generation HPE Networking Self-Driving solutions.
Pramod Badjate, President & GM, NETGEAR Enterprise, indicated that NETGEAR will develop new solutions using the switch technologies for high-bandwidth, intelligent, and secure enterprise applications.
Mark Gonikberg, senior vice president and general manager of Broadcom’s Wireless and Broadband Communications Division, said enterprises increasingly depend on AI for operations and security, requiring robust network infrastructure. He described the solution as providing a foundation for organizations to address risks in the AI era.
Siân Morgan, research director at Dell’Oro Group, commented that Broadcom’s end-to-end approach delivers enhanced performance, advanced analytics for AIOps, integrated edge AI, lower-cost MACsec security, and end-to-end telemetry.
