Benedict Evans called AI a "1997 internet moment."
He's right. But most growth teams are acting like it's 1994.
In 1997, the internet was real. Netscape had just gone public. You could feel the shift. But most companies were still treating "digital" as: build a website and see what happens. The actual value — rearchitecting how you acquire and retain customers at scale — came five years later, for the teams that understood distribution itself was changing. Not just content formats.
That's where we are with AI in growth right now.
I see a lot of teams adding AI to their existing workflow. Faster campaign briefs. Better copy variants. More A/B ideas per sprint.
That's 1994 energy. You've installed Netscape. You haven't thought about what the internet does to your distribution model.
The teams I think will win are asking a different question: which parts of our acquisition and retention logic actually break if AI is everywhere?
When every competitor can spin up 50 copy variants overnight, copy stops being a moat. When every user can get a personalized offer generated in real-time, generic lifecycle campaigns lose their edge. The loops that were defensible because they were operationally hard to execute — those are worth stress-testing now.
I'm not saying your entire growth stack needs to be rebuilt. I'm saying the teams waiting for the obvious signals have already lost the window.
1997 was obvious in hindsight. It wasn't obvious in 1997.