Cornered Data Owners
Willie Richter@willie
The idea
I believe that as AI gets cheap, the real edge shifts to whoever owns data nobody else can get, especially messy unstructured stuff like legal records, scientific archives, and real conversations, rather than generic content AI can generate itself. So this holds exclusive text archives, health and science records, data that gets licensed to AI labs, and locked-in anchors like credit and property data, alongside software that turns raw records into usable AI fuel. It's spread across five layers with a small cash reserve, built as a decade-long hold rather than a trade on any single name.
What it owns
- Software that organizes messy files, unstructured records and live data streams into formats AI can use generates revenue regardless of who owns the underlying data. · SnowflakeIt provides the cloud platform where companies store and prepare their raw data for AI applications, earning revenue from data handling rather than data ownership., MongoDBIt provides database software designed for unstructured data like documents, conversations and logs, the exact raw material needed to train and build AI applications., Palantir TechnologiesIt builds software that integrates scattered and messy records into coherent insights, increasingly powered by AI, for government and corporate customers., IBMIt owns software that streams and prepares live data across systems and its own AI tools, enabling customers to feed raw data into AI applications., ElasticIt makes search software that indexes huge volumes of unstructured text and logs, a common component in AI systems that need to retrieve information from messy data.29%
- Legal case law, scientific journals and professional research behind paywalls cannot be scraped by AI, forcing developers and institutions to license access instead. · RELX plcIt owns scientific journals through Elsevier and legal databases through LexisNexis, two of the largest locked archives that universities, law firms and AI developers must license., Thomson ReutersIt owns Westlaw, a comprehensive archive of case law and legal analysis that lawyers and legal AI tools cannot rebuild from free public sources., John Wiley & SonsIt publishes peer-reviewed academic journals and licenses that content to AI developers and research institutions, creating recurring licensing revenue.20%
- Insurance claims databases, credit ratings, property records and market indexes that entire industries depend on cannot be easily replicated, creating durable competitive moats. · Verisk AnalyticsIt holds decades of pooled insurance claims and risk data across the industry that individual insurers and outside competitors cannot easily recreate., S&P GlobalIt owns credit ratings, stock indexes and market data that financial institutions are required or expected to use, creating sticky long-term revenue., CoStar GroupIt maintains a proprietary database of commercial property details built over decades and cannot be assembled from public sources alone.17%
- Real human conversations, news archives and financial reporting are valuable training data that AI labs pay to license from the companies that own them. · RedditIt owns years of unscripted human conversations that AI labs pay to license for training, and the value grows as users continue posting new authentic content., The New York Times CompanyIt owns a large news archive and has demonstrated willingness to enforce licensing terms with AI companies, establishing that quality journalism text commands a price., News CorpIt owns Dow Jones, The Wall Street Journal and other news archives that it licenses to AI companies, showing that trusted financial content commands licensing fees.14%
- Clinical trial results, patient records and genetic data protected by privacy law are expensive to assemble and give their owners defensible advantages in pharma and biotech. · Tempus AIIt assembles linked clinical and genetic patient records that cannot be scraped from public sources, and drug makers pay for access to this proprietary dataset., IQVIA HoldingsIt holds large datasets of patient, prescription and clinical trial information that pharmaceutical companies depend on and cannot easily assemble elsewhere., Veeva SystemsIt operates the primary software platform where drug makers store clinical, regulatory and sales records, creating a steady stream of specialized pharma data.13%
- Cash7%