The Pattern Repeats
Everyone who called the internet a bubble in 1997 was right about the crash. And wrong about everything that mattered.
The NASDAQ gained 115% in the 36 months after ChatGPT's launch. It gained 110% in the same window after Netscape's IPO in 1995. The pattern match is not metaphorical. It is numerical. Cisco at its 2000 peak traded at 472× earnings with 17% net margins. Nvidia today trades at 56× with 50%+ net margins. These are not the same story.
In Vol. 15 — The Pattern Repeats, we run the actual data comparison between the internet cycle and the AI cycle — not the narrative version, but the numbers. $725B in hyperscaler capex in 2026 alone. 95% of enterprise AI deployments delivering zero measurable ROI. OpenAI at $25B+ run rate. The revenue-to-investment gap is real. So is the demand.
The question is not bubble or no bubble. It is which layer, which company, which valuation — and what signal tells you the thesis is broken.
The people who confuse those three things get both the bubble call and the opportunity call wrong simultaneously.
