Three ways to outperform: - Do way more than everyone else - Do way different - Do way less The last is easiest and most likely to win modestly and consistently, that makes it the hardest to do in a world that wants big upside fast.

Are we in an AI bubble?

It pays little to be skeptical early. It can pay a lot to be skeptical later as long as you weren’t skeptical early.

Most build AI to augment human work. Few build AI to replace work that can be augmented by humans.

Successful investing strategy goes with the nature of AI.

Everything’s a fantasy to those who can’t or won’t try to build it. Everything that’s come to life was once a fantasy for the person who built it.

Human labor is a guide to AI margins. AI apps that solve low intelligence problems will have low margins. Probably migrate to open source. AI apps that solve high intelligence problems will have high margins despite using sota models.

It's the part of the bubble where the proudest rational actors say its over. It's not over until those actors are absolutely besides themselves screaming that everyone else is irrational.

One thing AI won’t change - the power of a memorable idea and the words that convey it. GPT is one of those words that is inseparable from modern AI. It’s the foundation of foundation models. GPT is also now the ticker for Intelligent Alpha’s flagship Atlas ETF (formerly

“Why wont everyone just use AI to do that?” Anything that takes even minor effort won’t be done by most people.

Human slop is no better than AI slop.

2026 AI founders playbook: - Create games the ghosts can play well - Monitor and correct the ghosts as they play - Win