AI Bubble
Everything here starts on Nov 30 2022, the day ChatGPT launched. Every US-listed stock split into AI and the rest: prices, earnings and how much of each move came from a higher multiple. Then every AI category on its own, and what past bubbles did to investors who sat them out.
About this page
This page splits every US-listed company into the AI Total Market and the rest of the market, then follows each group's price, trailing earnings and price to earnings since ChatGPT launched on Nov 30 2022. Each group's price move is split into earnings growth, the change in its multiple and dividends, the same arithmetic as Return Drivers.
The AI list is curated: a company joins when AI products or AI demand are its largest growth driver in its own filings, from chip makers and the hyperscalers that build the data centers to the power producers that feed them and the software companies selling AI features. Every AI figure is shown twice, with and without Microsoft, Amazon, Alphabet and Meta, because AI is a slice of much bigger businesses for those four.
The last sections ask what past bubbles did to investors who sat out the bubble sector, held the whole market or held only the bubble sector, using a century of industry returns (radio and electricity in the 1920s, oil around 1980, the dot-com years and the housing boom), and compare the earnings behind each run-up with the earnings behind AI today.
What counts as an AI stock here?
A company whose own filings name AI products or AI demand as its largest growth driver. The list is reviewed each quarter and every change is dated. Apple is left out because AI is not its growth driver in its filings; Tesla is in because its valuation rests on self driving and robots.
What is multiple expansion?
The market paying more per dollar of earnings: the P/E rising. A price move is earnings growth times the change in the multiple, so the page shows how much of each group's move came from each.
Does the whole market fall when a bubble sector bursts?
The Past bubbles table answers from the data: what the rest of the market did in each bust, where it stood two years after the low, and how long it took to regain its earlier high.