For over a decade, the narrative around IoT has been straightforward: data is valuable. Connectivity was the plumbing—important, but secondary. The real action happened when that data reached the analytics dashboard, the operations centre, or the business intelligence system. That story is changing fundamentally. Today, powering AI-driven outcomes are becoming an increasingly important part of the IoT data story. And connectivity’s role has shifted from simply transporting data to enabling faster, smarter, real-time decision-making. This shift is forcing a reckoning with how organisations manage their IoT operations—and it’s moving well beyond the portal. The network layer has transformed Start with what’s happening at the infrastructure level. A few years ago, there wasn’t much choice: you got a SIM card, a 3G or 4G connection, and you figured it out. Now, dedicated IoT networks like NB-IoT and LoRaWAN optimise for low power and broad coverage. 5G provides ultra-high capacity. Satellite connectivity enables remote deployments. Standards like GSMA’s SGP.32 are also enabling more flexible remote provisioning and management of IoT connectivity. Connectivity platforms can increasingly automate actions such as network selection, profile management and connectivity policies based on changing conditions. Connectivity is no longer binary or static. It’s intelligent, adaptive, and increasingly automated. The application layer is where things get interesting But another important transformation is happening one layer up, in how IoT data flows into business operations. Historically, application enablement meant deploying dashboards for people to monitor and interpret data. The portal was the centrepiece—the place where decisions happened. That model is breaking down, and for a good reason: it’s bottlenecking decision-making. Today’s developer tools, APIs, and automation frameworks bypass the dashboard entirely. IoT data routes directly to where decisions are actually made—ERP systems, operating systems, AI agents, and business processes. Data no longer needs human interpretation in a portal. It