Most companies rolling out AI right now are about to get beat by the ones who...

Most companies rolling out AI right now are about to get beat by the ones who waited.

In the 1880s, factories swapped steam engines for electric motors and kept everything else. Same floor layout. Same processes. Same walls. Productivity barely moved for 30 years.

Worse: the factories that electrified first lost ground to competitors who waited. The early movers kept the old factory and plugged in a new power source.

Diana Mounter quoted George Sivulka on this at GROK Conf this morning: "We swapped the motor. We have not yet redesigned the factory." That's where almost every company is with AI right now.

Peter Barth spoke earlier at the same conference. He spends his days with executives at mid-size and enterprise companies, and his observation from the last six months is blunt: roughly 80% of them have bought something with "AI" in the name. Adoption is near-universal. Impact is close to zero.

The reason shows up in the conversations. A exec tells him, "My mandate is to bring in an AI company." He asks what they want to do with it. The answer: "Literally, no idea." The mandate came from people who've never used the tool. The people who'll have to use it weren't asked what it's for. The strategy document says "AI adoption." The adoption happens. Nothing changes.

That's an electrified mill with the old floor plan.

If you're running a company right now, here's the question I'd sit with (and I mean actually sit with it, not hand it to a consultant). Has anyone on your leadership team used the thing for more than fifteen minutes? Not a demo. Not a vendor deck. Fifteen minutes of real work on real inputs. If the answer is no, you don't have an AI strategy. You have a purchase order and a hope.

The factories that won didn't have better motors. They had leaders who understood what the new power source made possible and redesigned the work around it. That takes time, it costs money, and it doesn't look like a press release.

It's also the only thing that has ever actually worked.