Lao Huang says AI is a new industrial revolution, and I think he's half right.

The part that's right: The whole world is scrambling for GPUs, just like scrambling for oil back in the day. Training a large model burns enough electricity to power a small town. Big tech companies are lining up to send money to Nvidia, and some smaller companies are even paying in full half a year in advance just to get in line. The hype is real.

But to be blunt—what are you using AI for right now? Writing weekly reports, editing photos, chatting to kill time? Where's the world-changing disruption everyone promised? Designers are still working overtime, programmers are still fixing bugs. Basically, AI has only replaced jobs that were already becoming obsolete anyway.

Lao Huang obviously has to hype it up hard. He's selling shovels, and the crazier the miners get, the more money he makes. Nvidia's market cap hit $2 trillion, more than all other chip companies combined. That valuation is betting on one thing: everyone will keep buying GPUs forever, without stopping.

But history keeps teaching us the same lesson—pigs can fly in a bubble, but when the wind stops, they crash the hardest. The fiber optic bubble in 2000 laid down tons of cables nobody ever used. What if we're laying down tons of computing power nobody ever uses?

That said, I'll give AI this: the iteration speed is genuinely insane. Two years ago when ChatGPT came out it looked like a toy, and now large models can run on phones. At this pace, who could possibly say what it'll be like in three years.

So the revolution might be real, but we're probably still in 1780, right when the steam engine was just invented, way before trains started running all over the world.

Don't rush to go all in, and don't rush to write it off either.