Ask not what AI can do for Wi-Fi, but what Wi-Fi can do for AI

Exec Talk article by Shahnawaz Siraj, Director of Technology | Intelligent Connectivity at MediaTek

AI is transforming industries, and without exception, it has become the buzzword of networking. While vendors and equipment manufacturers are pitching the same AI use cases for networking, such as configuration management, self-driving networks, anomaly detection, application visibility and QoS, and location services, the real differentiation in the execution - how reliably and at scale these solutions work.

On end-user devices, the role of AI is obvious: live translation, image editing, and voice assistants. In the cloud, it’s even clearer: the cloud is the “ocean” where all the data flows, it is where centralized AI workloads can see the big picture, correlating patterns across millions of devices and networks. That’s what enables use cases like anomaly detection and others.

The harder question has been how AI is reshaping the edge, and in particular, the Wi-Fi edge.

In the home, the gateway is the logical edge compute point before traffic heads to the cloud, especially now that integrated Wi-Fi has become the standard.

When it comes to the enterprise, the definition of “edge” can be less clear: is it the WAN edge (SD-WAN/secure gateway), or the Wi-Fi access points where client devices connect? If we extend the home analogy, the WAN edge remains the logical place for heavier AI workloads.

It best addresses the key concerns of data immediacy, real-time responsiveness, latency, privacy, and network bandwidth efficiency.

Which then raises the harder question: what’s the unique role of APs (Access Points) in AI at the edge? 

The reality is that the industry hasn’t yet landed on a breakthrough “AI-on-AP” use case, and that’s fine. The role of the AP isn’t limited to running AI itself, but also to be ready for the traffic shifts AI will create, through programmability, better QoS, and more flexible platforms. That’s where APs can make a real impact and this is where the conversation turns 180 degrees. It’s not just AI for Wi-Fi, but also Wi-Fi for AI.

Are Wi-Fi networks ready for the new, unpredictable traffic patterns that AI applications will generate? End devices will keep getting smarter and running more inference locally, but that won’t reduce the demand on networks. Instead, it will create new and unpredictable patterns: burst of high-throughput or latency-sensitive traffic and dynamic flows across devices and the cloud. That means Wi-Fi’s role is not just more bandwidth, but resilience, adaptability, and future-proofing against traffic we can’t yet define today.

The industry is also moving to standardize this space. Within IEEE 802.11, the AI/ML group is set to explore how AI can be embedded into Wi-Fi. There are use cases being discussed to enhance WLAN performance, for example, the AP can leverage dynamic capabilities of AI/ML to assist STAs in proactively selecting optimal configuration parameters. In parallel, there are use cases focused on enabling AL/ML itself over WLAN, for example, supporting federated learning across Wi-Fi networks.

At MediaTek, as an innovation and technology leader, we see ourselves as enablers. We develop and provide the complete platform, hardware and software together, along with the SDKs, tools, and reference use cases that let our partners and customers innovate further. Whether it’s optimizing QoS using AI, enabling hooks for edge analytics and troubleshooting, or preparing APs for workloads that haven’t even been imagined yet, our goal is to make sure Wi-Fi isn’t just keeping up with AI, but empowering it.

The future of AI will depend on the networks that carry it. And Wi-Fi, more than ever, is at the front line.