**TLDR:** The next phase of the AI boom may shift from compute & infrastructure toward companies that own continuously updated proprietary datasets, workflows, and human interaction data that frontier models depend on.
**Positions**:
RDDT $300 Sept 2026
TEM $120 Jan 2027
\-
We are coming into the home stretch of the initial AI supercycle boom, and the market has its obvious winners: chips, compute, power, memory.
The next major AI bottleneck may be proprietary data.
AI models are only as valuable as the data they can continuously access, update, and train on. The next wave of AI winners won’t just be infrastructure companies, they’ll be the platforms that OWN the datasets, data workflows, & data APIs. This is the shit everything is built on top of - THE OIL TO THE ENGINE
Healthcare is a clear example. Medical knowledge changes constantly, and models can’t rely on static scraped internet data. Companies with continuously updated clinical databases, workflows, and physician interactions become mission critical training and inference layers.
There’s also human conversation & behavioral data: dialogue, emotion, engagement, intent. That's a premium not only for Dead Internet Theory , but also fuel for frontier models. Platforms once valued as “social media” or “user-generated content” may instead be valued as DATA RESERVOIRS for AI ecosystems. Reddit is valued at less than 2% the size of META - still apples to apples (social media -> users -> engagement -> ads -> revenue). No data reservoir boost
Data commoditization is nothing new but the last few years have been the Wild West. And CEOs are now starting to openly discuss data licensing, API monetization, and AI partnerships as CORE business lines. Tempus CEO Eric Lefkofsky highlighted the data business as a multi billion dollar opportunity. Reddit CEO Steve Huffman calls user content "modern oil" for AI.
The market still largely values these companies like traditional software, media, or database businesses - not strategic AI data suppliers.
The AI infrastructure trade may be crowded. The AI data layer trade is just beginning.