I'm a fan of Reuter News, and a few years ago I looked up the company and realised that they are mainly a legal and accounting technology company. They have always been trading at a hefty premium, until the Saaspocalypse, where I watched them getting sold off into oblivion.
Here I am going to argue that unlike other saaspocalypse victim, Thomson has a significant moat under its 150-year-old Westlaw database, that spans more than 100k subtopics. This is a manually curated database, meaning their legal professionals has been inputting data and annotation one by one into the database over the years. This data is also proprietary, meaning no one else have access nor can train their AI with the Westlaw database.
The CEO Steve Hasker has pointed on multiple earnings call and interview that in legal and accounting: reliable and verified answer/data is critically important. Therefore, even a small occurrences of AI hallucinations is not acceptable and can prove to be fatal.
Still, management knew what AI can achieve, and has been actively adopting AI, acquiring another company to release the CoCounsel Legal tool: which has been recently rebuilt on top of the Anthropic Cloud Agent SDK.
So, the AI disruption narrative relies on Anthropic (and other AI chatbots/wrappers) usage for legal and accounting purposes, making it cheap and reliable enough that customers need less of Thomson's tools and services. But for enterprises, I think it is simply not good enough to just use Anthropic Legal Plugin. They need to be able to accurately verify the data, know where its coming from, and point to the source of truth.
Thomson Reuters is aiming to offer this exact layer which integrate the power of AI with their century old legal database, allowing for authoritative data: verified, with clear source of truth.
The company now trades at around 27x trailing earnings, which is not cheap in pure number, but given that they have a durable competitive advantage, I believe this is undervalued for the company. Their average PE historically is also closer to 35x.
I think the Westlaw database is an incredibly wide moat and a significant asset, which should allow them to build an AI tool that outperforms generic AIs and other fine-tuned GPT-wrapper.
Thoughts?
My full analysis here: https://open.substack.com/pub/economiyaki/p/thomson-reuterss-150-year-old-westlaw?utm\_source=share&utm\_medium=android&r=2wzuop
