AI stocks climbed to valuations that assumed companies would still be printing money when your great-grandchildren are comparing holograms of their dead pets. The entire trade rested on one premise: earnings so obscenely high and so hilariously distant that nobody could disprove them with a calculator.
Turns out discount rates exist.
Rising bond yields do one thing well. They make future cash flows worth less today. The higher the yield, the less you pay now for a promise twenty years out. AI companies promised ultrahigh earnings in 2040. Investors paid 2040 prices in 2024. Yields went up. Math went to work. Stocks went down.
Retail traders bought the top because they read that AI would replace every job except theirs. They watched Nvidia quintuple and assumed the next semiconductor company with a three-letter ticker would do the same. They did not consider that a 10-year Treasury yielding five percent makes waiting for robot profits in 2043 feel less appealing than it did when bonds paid two percent.
The bubble question misses the point. Every stock is a bubble if you squint hard enough at the assumptions. AI stocks just made it easier. They offered no earnings today, vague promises tomorrow, and valuations that required you to believe in both compound growth and the heat death of the universe arriving late.
Yields rose. The present value ofζ°ΈθΏ collapsed. Portfolio managers who bought AI stocks at sixty times sales suddenly remembered they had risk committees. The committees asked one question: why are we paying for earnings in 2045 when we can get five percent today from a government that prints its own money?
Nobody had a good answer. They sold. The stocks fell. The headlines wrote themselves.
Journalists now ask if rising yields could pop the AI bubble, as if the bubble needed help popping instead of just time and a basic understanding of how present value works.
Photo by Google DeepMind on Unsplash
Leave a Comment