You Installed the Electric Motor. You Forgot to Redesign the Factory.

The electric motor already existed in 1880. It worked, it was reliable, and any factory owner could buy one. And yet, by 1900, less than 5% of the motive power in American factories came from electricity. Almost everything still ran on steam.

Why did something already ready for use take so long?

Stanford economist Paul David asked himself that same question in 1990, in a study that is now a classic in the economics of technology. What he found was this: the first factories that installed electric motors changed nothing else. They kept the same design they used with steam — a giant central shaft running the length of the plant, connected by pulleys and belts to each machine. They simply replaced the steam boiler with an electric motor at the end of that same shaft. The technology had changed. The factory had not.

The problem is that shaft-centered design existed for a specific reason: steam could only be distributed mechanically, so everything had to be physically connected to a single power source. Electricity had no such constraint. It could be sent to each machine separately, with its own small motor, without needing a central shaft. But understanding that — and redesigning the entire factory around that idea — took the industry almost three decades. Only in the 1920s, when a new generation of plant managers built factories with an individual motor on each machine, did productivity explode. Floor layouts were no longer designed around where the shaft had to run. They were designed around where work had to flow.

Almost forty years between the motor being available and the factory finally being redesigned to take advantage of it.

Now replace “electric motor” with “artificial intelligence” and you’ll recognize exactly the moment where your company is stuck today.

Most organizations have already installed the motor. They have Copilot, they have ChatGPT, they have some assistant connected to the CRM. But everything else still runs on the same old central shaft: the same approval committees, the same seven meetings to make a decision, the same processes designed for a world where generating a report, a proposal, or an analysis took days. You installed an electric motor in a steam factory and wonder why productivity isn’t moving.

This isn’t an accident or an exception. It is, in fact, the pattern that all general purpose technologies follow — that same concept we discussed last week when everyone was looking at the wrong GPT. Electricity, the internet, and now artificial intelligence all move through the same arc in three stages. First, a phase of enthusiasm and laboratory experimentation, where the technology exists but lives isolated in specific projects. Then, a long and confusing stage of complementary investment, where organizations must redesign processes, train people, and reorganize work — and where results are slow to appear because the costs of that redesign arrive before the benefits. And finally, permeation: the technology becomes so embedded in how everything operates that it stops being talked about as something special.

Electricity has completed that arc. So has the internet. Artificial intelligence is today somewhere in that second stage: the one where the central shaft still hasn’t been dismantled.

The good news is that stage, confusing as it is, isn’t a dead end. It is, historically, the exact place where advantage is built: while most people keep looking at the motor wondering why it isn’t performing, those who understand the pattern are already dismantling the shaft.

Do you already have the motor installed in your company? That’s the easy part. The question that really matters is whether you’ve started redesigning the factory floor.