What will separate AI leaders from AI followers in 2026? 

The companies pulling ahead have stopped treating AI as a productivity tool and started treating it as a core capability. What I see working with industrial companies is that followers run pilots that never scale, because the use case was never a real business initiative. Leaders do the opposite: they identify the one problem that, if solved with AI, changes a core metric and commit to it at the strategic level. The second separator is model ownership. Companies running AI on sovereign infrastructure, trained on their own proprietary data, are building competitive advantage their rivals simply cannot replicate off the shelf. That gap compounds with every passing month, and catching up later will be far harder than it looks today.

Why do so many AI projects still struggle to deliver measurable business value? 

The core problem is that most companies are running AI as a technology programme when it needs to be a business programme. You end up with impressive demos, enthusiastic IT teams, and KPIs that measure adoption rather than outcomes. The second issue is that generic AI cannot know what makes your organisation distinctive – your proprietary processes, your engineering tolerances, your decades of accumulated operational data – and that is precisely where business value lives. Companies that close that gap, by deploying AI trained on their own data and embedded in their actual workflows, see results that compound over time. Those that keep experimenting at the edges will keep getting edge-case results. The question every leadership team should be asking is not "where can we use AI?" but "which one problem, if solved, would change a number that matters to us?" and then commit to that fully before moving on to the next.