Paul Viorel

Industrial environments have been built one use case at a time, leaving most large manufacturers with fragmented IT and OT infrastructure. A manufacturer with 10 sites might run four different home-grown MES platforms, three different ERPs, and entirely custom data connections between them. As organizations deploy artificial intelligence, that absence of standardized data architecture has become the primary bottleneck. AI hits a wall when data is uncontextualized.

Close to 80% of power generation deployments in ABB’s global electrification and power install base run on-premise at the edge. That figure, shared by Cody Falcon of ABB Energy Industries during IIoT World Energy Day 2026, is not an aspiration. It is the current state of how power gets managed. The reasons are practical: battery systems that must respond in 150 milliseconds, air-gapped sites with no cloud connectivity, and data volumes growing too fast and too expensively to ship off-site.

In manufacturing, the window between a small deviation and a costly event can be measured in seconds. A vibration pattern shifts, a batch parameter drifts, a temperature moves outside its normal band. The data exists, but by the time it reaches the person who can act, the window has closed. Response speed for AI in manufacturing is a technical challenge and a management challenge in equal measure: who sees the signal, who trusts it, and how fast the organization responds.