Does AI in 2025 feel like the Internet in 1999?

These days with AI and LLMs feels like feel like Internet circa 1999. Agree?

In 1999, I was in my Senior year at Queen’s University, completing my Engineering degree.

I wasn’t an investor at the time, but I followed the markets a bit since I was starting to look into employment opportunities. I remember a distinct thought from back then: “Why am I learning C++? Every ‘hot’ job wants Java developers?”.

Anything internet related or with .com in the name was getting funded or and valuations were going to the moon. You have to understand the times: my university email was my first email address in 1996, and we were accessing it through Pine. (Sidenote: Hotmail and Yahoo mail only started in 1996 and 1997, respectively)

These days, AI seems to be developing as fast, if not faster, than the move to the internet. Every week, it seems like one model is leap frogging the other. Even developing new LLMs seems to be faster/cheaper because those models can be developed by training/prompting on existing LLMs.

MCP was introduced by Anthropic in November 2024 and it’s everywhere already. It now feels like a race to build “Whatever” (pick-your-niche) you can find and incorporate AI into it using the APIs. Just like there was a race to [English-word].com in 1999.

Don’t get me wrong, some amazing things have been built. But I’m not seeing any moat on so many of these businesses. It’s the AI wild west gold rush, just like it was the internet gold rush in the late 90’s.

All I know at this point is that this AI train is a major train, like the internet was over the last 25-30 years. Don’t sleep on it. Millions will be made, millions will be lost. But this will change our lives in the years to come.