Net Data Creators

There’s one question that cuts through everything in AI investing right now: does this company create data, or does it consume it?

A net data creator generates proprietary information as a byproduct of operating its core business. Google processes every search. Visa processes every transaction. CrowdStrike’s sensors sit on hundreds of millions of endpoints. NVIDIA’s CUDA ecosystem produces developer telemetry no competitor can replicate without rebuilding the entire developer base. Meta’s social graph and Palantir’s customer ontologies are built transaction by transaction, interaction by interaction, over years. The data can’t be scraped. It can’t be licensed. It doesn’t exist anywhere else.

A net data consumer uses data rather than creating it. Salesforce runs on records customers upload. Adobe’s generative tools were trained on licensed corpora. ServiceNow, Workday, HubSpot, and Atlassian have real moats in workflow, distribution, and switching costs, but not data moats. A sufficiently capable frontier model with an API key can do what many of them do.

The chart shows what three years of that distinction looks like in the market. An equal-weight basket of 44 net data creators, from Google and Visa to CrowdStrike, Deere, Intuitive Surgical, and MercadoLibre, returned +144% since May 2023. An equal-weight basket of 18 net data consumers, including Salesforce, Adobe, ServiceNow, Workday, HubSpot, Atlassian, Oracle, SAP, and others, returned +21%.