By 2036, Saying “We Have an AI Strategy” Will Sound as Strange as Saying “We Have an Internet Strategy” Does Today.

I want to make a prediction.

By 2036, saying in a board meeting “we have an AI strategy” will sound exactly like saying “we have an internet strategy” sounds today. Almost embarrassing. Evidence that at some point you treated a general purpose infrastructure as if it were a specialty.

Nobody today says they have an internet strategy. The internet is, simply, how everything works: how you communicate, how you serve customers, how your supply chain runs, how your team coordinates. It’s invisible because it’s everywhere.

That’s where artificial intelligence is headed. Not as a department, not as a roadmap, not as a named strategic initiative. As the operating system for how decisions are made — woven inside every function, invisible precisely because it’s taken for granted.

We’ve spent ten columns building this argument, and it’s worth walking through the full journey one last time, because each piece only makes complete sense alongside the others.

We started with IBM and its 1997 “e-business” campaign — half a billion dollars to convince the world that the internet needed its own department — and with the joke that ended up being the thread running through the entire series: there are two GPTs, the one everyone chases (the model, the chatbot, the launch of the month), and the one that actually explains everything happening (general purpose technology). We moved to Walmart, which gave the internet its own CEO in 2000 and took sixteen years to undo that separation. Then we went even further back, to 1880, to show that the electric motor was available for nearly forty years before anyone dared to redesign the entire factory around it — and why that explains that your AI ROI probably isn’t behind schedule, but exactly where economic history says it should be.

Then came the real urgency: AI reached 1.2 billion users in under three years, something that took the internet nearly a decade — which means the fifteen-to-twenty-year mental calendar many companies still operate on no longer matches the actual pace of this change. And alongside the urgency, the hope: Amazon was a garage startup in 1994 facing a Walmart that outranked it on every metric, and thirty-two years later surpassed it as the largest company in the world, because it built everything around the internet instead of bolting it onto what it already had.

We talked about the difference between having a chatbot — which is like having a website in 1999 — and having AI truly integrated into how decisions are made, the only thing McKinsey found that predicts whether a company sees real impact on its results. We saw that IBM, the same company that wrote the separate-department manual in 1997, today has no Chief AI Officer, and instead chose to distribute that responsibility among every area leader. And we closed last week with the most important permission of the entire series: imperfect AI, integrated today into a real operation, will almost always beat the perfect AI that someone is still planning for next year — just as Amazon, eBay, and Google never waited for broadband.

Ten columns, one argument: artificial intelligence is a general purpose technology, and the window to build advantage before it normalizes is open right now.

The leaders who win this transition won’t be the ones with the best AI roadmap. They’ll be the ones who stopped needing one — because they rebuilt everything around AI long before it became obvious to everyone else.

What will sound stranger inside your company ten years from now: having waited too long, or having started too soon?