Proprietary Data Compounders
Oren J. Falkowitz@orenfalkowitz
+0.9%
since Oct 6
The idea
I believe AI models are becoming commoditized, so the real edge shifts to companies sitting on continuous, non-scrapeable first-party data they can turn into compounding product advantages. This take leans heaviest into physical-world telemetry, hardware-tied data like fleets, equipment, and sensors that can't exist without the installed base, then rounds out with biometric data owners, transaction-intent platforms, and enterprise workflow systems. It's built as a multi-year structural hold across those pillars with a modest cash reserve, and no single name dominates the bet.
What it owns
- Data generated by installed machines and sensors in the real world cannot be scraped or replicated, creating a durable competitive moat. · Deere & CompanyConnected farm equipment captures field-level soil, yield, and machine performance data that no model can access without hardware installed on working tractors., Tesla, Inc.Tesla's fleet records billions of real driving miles that exist only because the hardware is deployed on roads, generating telemetry no competitor can scrape or replicate., Samsara Inc.Sensors inside commercial trucks and worksites generate operational data that cannot exist without the company's hardware installed in the field., The Progressive CorporationYears of permissioned driving-behavior records from its telematics program let it price risk more precisely than insurers without access to that data., Axon Enterprise, Inc.Body cameras produce evidence footage behind an authenticated system no one else can access, the clearest case of proprietary hardware gating mission-critical data., Caterpillar Inc.Over a million connected machines send usage and condition data from mines and job sites, telemetry that cannot be replicated without the installed iron., Trimble Inc.Positioning equipment collects centimeter-level geospatial data from construction sites and farms, physical telemetry tied to instruments deployed in the field.41%
- Domain-specific logs from mission-critical software compound into AI features competitors without the install base cannot match. · ServiceNow, Inc.The platform records how work flows through large organizations, tickets, approvals, incidents, creating workflow data that compounds inside its own walls., Veeva Systems Inc.The system holds regulated clinical trial and doctor communication data for drug companies, records that cannot be scraped from outside the platform., Salesforce, Inc.The platform stores decades of customer interaction records from the largest sales organizations on earth, data that compounds through internal AI development., Datadog, Inc.The platform ingests live telemetry of how modern software actually runs, data customers generate continuously and that feeds its own AI monitoring features.20%
- Continuous health and body data sits behind proprietary hardware and user permission, so no outside AI model can access it. · Apple Inc.The Apple Watch collects continuous heart, sleep, and activity data behind proprietary hardware and user permission, locking biometric signals inside the ecosystem., Garmin Ltd.Deep training and recovery data from serious athletes who wear its devices around the clock stay locked inside the ecosystem through continuous performance insights., DexCom, Inc.Sensors generate round-the-clock glucose readings, medical-grade telemetry that only exists on its hardware and behind patient consent.16%
- Verified purchase and ledger data shows what people actually do with money, a signal no public dataset contains. · Intuit Inc.The platform sits on the actual books of millions of small businesses, holding verified financial data that reflects real activity rather than stated intent., Visa Inc.The network observes verified purchase behavior at a scale no scraped dataset can match, capturing transaction signals across the global financial system., Toast, Inc.The platform captures item-level restaurant transactions through terminals it installed at the point of sale, demand data that exists nowhere else.15%
- Cash8%