Every few decades a technology shows up that doesn’t just add a feature to how we live, it replaces the operating layer underneath it. In the 90s, that was the internet. Right now, it’s AI. And the parallels aren’t just convenient marketing copy; they’re structural.

The internet’s core trade was access for effort. It put nearly all human knowledge within reach, but you still did the work, searching, filtering, reading, deciding. AI removes that step. You don’t find information anymore; you ask for the answer. You don’t interpret a spreadsheet; you ask what it means. The shift from search engines to reasoning engines is as big as the shift from card catalogs to search engines was, and it happened in about three years instead of three decades.
The interface has changed, not just the tool. A decade ago, “using software” meant clicking through menus designed by someone else’s assumptions about your workflow. Now it increasingly means describing what you want and watching it happen: “cut this video down to the highlights,” “build me a feature that does X,” “lay out these slides.” Language is becoming the interface, which means the companies that win aren’t necessarily the ones with the best buttons, they’re the ones whose product understands intent. That’s a genuinely uncomfortable position for software incumbents who spent twenty years optimizing UI, not understanding.
This is also why the comparison to early internet media disruption is more than a rhetorical flourish. Newspapers in 1998 didn’t think Craigslist was a threat to classified ads, it was just a website. Most software categories today are making the same miscalculation about AI: treating it as a feature to bolt onto the existing product instead of recognizing it as a different kind of product entirely.
A new economy is forming underneath the old one, the same way e-commerce and cloud computing formed underneath the early web. AI-native productivity tools, autonomous agents, synthetic media, AI-assisted scientific research, personalized education, these aren’t incremental improvements to existing categories. They’re categories that didn’t exist five years ago. And if the internet’s adoption curve is any guide, the tools available right now are the dial-up version of what’s coming: today’s chatbot interfaces will look as primitive in five years as 56k modems look now.
The part that matters most for individuals, though, isn’t infrastructure, it’s access. The internet’s biggest social effect wasn’t commerce; it was that it let anyone publish, with no permission required. AI does the same thing one level up: it lets anyone produce. You no longer need design training to make something visually competent, coding fluency to ship a working prototype, or domain expertise to make sense of a dataset. That doesn’t replace expertise, it changes who gets to start.
So yes, AI is the new internet, not because it competes with it, but because it’s the next layer built on top of it. The internet connected the world’s information. AI is starting to make that information actionable on demand. The people who treated the early internet as a toy in 1996 spent the next decade catching up to the people who treated it as infrastructure. The same choice is in front of everyone again, except this time the window to get ahead of it is shorter.

There’s a piece of biology worth borrowing here. A mushroom never grows alone, it’s the visible fruit of a mycelium network that’s been spreading underground for years, invisible and patient, doing the real work long before anything breaks the surface. The internet was that mycelium: decades of infrastructure and connection growing quietly until the conditions were finally right. What we’re calling AI is the mushroom, sudden, exposed, and already decomposing the old structures to feed whatever grows next. And like any fungus, it doesn’t ask permission before it spreads. The part people forget is that a mushroom is never the end of the cycle, its spores are already becoming the mycelium for whatever fruits after it. AI won’t be the last mushroom either.


Leave a comment