AI Reached 1.2 Billion Users in Under Three Years. It Took the Internet a Decade.

More than 1.2 billion people have already used some form of artificial intelligence tool, according to Microsoft’s latest AI Diffusion Report, published this year. Getting there took AI less than three years. For the internet to reach a similar number of users took roughly a decade. For electricity, it took even longer.

The report itself frames it in a way that connects directly to everything we’ve been discussing in this series: AI is a general purpose technology that stands on the shoulders of three predecessors — electricity, connectivity, and computing. It didn’t invent itself from scratch. It inherited them. And that’s why it moves much faster than any of its predecessors.

Here’s the concrete reason. The internet in the ’90s needed to build its own physical infrastructure from scratch: cables, servers, home networks, access providers. Each new user meant laying new cable. Artificial intelligence doesn’t have that problem. It runs on the cloud, on the internet, and on data centers that were already built. It doesn’t need to lay new rails. The rails were already there, and AI simply climbed on top.

That completely changes the timeline calculation that most companies still have in their heads. If the internet took fifteen to twenty years to go from “we have an internet team” to “everything runs online,” the implicit assumption in many boardrooms is that AI will follow a similar calendar — and that there’s plenty of time to define a strategy at a measured pace. The numbers say otherwise. Every “we’ll deal with that later” decision is being made today on an assumption that’s no longer true.

And this speed isn’t even. The same Microsoft data shows that countries like the UAE or Singapore already surpass 60% of their working-age population using AI regularly. In parts of Latin America, that number is still below 10%. For a company in Chile or anywhere in the region, this isn’t just a race against the 1990s internet timeline. It’s a race against a regional average that’s already running slower than the rest of the world — which means every year of indecision weighs twice as much.

I’m not writing this to generate panic. Panic leads to bad decisions, just like complacency does. What I do believe is that it’s worth recalibrating urgency to match the real speed of this transition, instead of operating with the stopwatch of a previous technology wave. If your company is following the internet playbook — “let’s build a team, test a few use cases, see how it goes” — you’re probably already further behind than you think relative to the actual pace of this change.

The good news is that understanding the right speed is also an advantage. Companies that have already adjusted their internal clock to this new timescale don’t need to rush desperately. They just need to stop planning as if they have twenty years ahead of them, when the evidence says they have considerably fewer.

Is your organization making AI decisions with the 1990s internet stopwatch, or with the actual 2026 stopwatch?