116 THE LAST WORD The agentic payoff Manufacturers are no longer experimenting with AI. Instead, they are deploying agentic operations powered by data platforms and sustainable computing When the manufacturing industry comes together at Hannover Messe on 20-24 April 2026, the atmosphere will be markedly different from the AI tourism that has defined the last two years. If 2024 was the party, 2025 was the renovation – a year of hard, unglamorous work where the industry shifted focus to ‘the year of the platform’. We have spent the last 12 months breaking down silos, deploying industrial data fabrics and wrestling with the productivity crisis that saw post-pandemic gains evaporate. The scale of the challenge has been, and still is, enormous. For example, food and beverage corporation PepsiCo manages around 60 petabytes of enterprise data across 200 countries, 291 manufacturing sites and more than one billion consumer touchpoints every day. But foundational work in data has set the stage for a thrilling payoff in 2026: the era of agentic AI. The top 13 per cent of industrial organisations have stopped treating AI as a siloed IT experiment and have weaponised it as a core strategy, we found in our 2026 Industrial AI Pacesetters Report. At this year’s fair, the ‘Industrial AI in Action’ theme championed by Microsoft and its ecosystem is no longer aspirational – it is operational. We are moving from chatbots that summarise text to autonomous agents that reason, plan and execute changes in the physical world. However, this shift brings a new challenge called the ‘agent lake’. As vendors like Siemens and Rockwell Automation deploy specialised agents, the risk of uncoordinated bot-on-bot conflict increases. The priority is no longer just adoption, but orchestration – governing these agents to ensure they negotiate, rather than collide, over production schedules and energy assets. Optimism is warranted because the sustainability shift has finally arrived. Hannover Messe’s partner country, Brazil, will showcase how sustainable computing can turn the energy-intensive training of these agents from a liability into a decarbonisation strategy. This geopatriation of data is a pragmatic solution to the sovereignty and sustainability demands of the European market. To get beyond the hype, look for the ‘boring’ wins on the show floor. Ignore the flashy demonstrations and look for the semantic layers and context engineering that prevent AI hallucinations. The winners of 2026 are those who used 2025 to decouple their data from their applications, creating a sovereign ground truth. The plumbing is done. The platforms are ready. It is time to let the agents work. Colin Masson is director of research at ARC Advisory Group COLIN MASSON: ARC ADVISORY GROUP
RkJQdWJsaXNoZXIy NzQ1NTk=