Sarat Sethi thinks AI fears tanked Verisk Analytics enough to make it worth buying. The fears being that artificial intelligence might somehow compete with a company that collects insurance data. The data nobody else has. The data Verisk spent decades accumulating while everyone else was doing literally anything more interesting.
Verisk's business model works like this: insurance companies need data to price risk. Verisk has that data. Verisk sells access to that data. Recurring revenue streams in because guess what, insurance companies still need the data next quarter. Hard to replicate means competitors would need to spend thirty years building the same datasets, which they won't, because they have shareholders who enjoy living.
But some retail trader read that ChatGPT can write a haiku about flood insurance and decided Verisk was finished. Sold at a loss. Posted about it in a Discord. Three hundred other geniuses saw the post and sold too. Stock dropped. Sethi looked at the chart, saw the dip, checked if the actual business changed. It hadn't. He recommended buying.
The AI fear trade relies on believing that language models will spontaneously generate proprietary insurance loss databases from thin air. They will not. They can summarize existing data. They can generate text. They cannot materialize information that does not exist in their training sets. Verisk owns information that exists nowhere else. This is not complicated.
Technical analysts don't care why stocks move. Retail traders panic-sold a data monopoly because they confused software that writes emails with software that replaces irreplaceable datasets. The price dropped below value. Sethi saw the gap. He called it a buying opportunity.
Somewhere right now a day trader is Googling whether AI can disrupt recurring revenue models, getting confused by the results, and deciding to buy Tesla calls instead.
Photo by Kanchanara on Unsplash
Leave a Comment