The core names discussed by the OP are Reddit (RDDT) and Tempus AI (TEM). The thesis is that the next phase of the AI trade may rotate from infrastructure such as chips, compute, power supply, and memory into companies that control continuously updated proprietary data, workflows, and API gateways. The OP's positions are RDDT September 2026 $300 calls and TEM January 2027 $120 calls, implying the view that neither company has yet been fully re-rated as an “AI data-layer supplier.” The OP’s evidence chain has two parts. First, medical AI cannot rely only on static internet text, because clinical knowledge, drug development, patient journeys, and physician interactions are constantly changing. Tempus owns multimodal healthcare data spanning clinical, genomic, imaging, and real-world evidence data, and monetizes it through data licensing to pharmaceutical companies, modeling, and analytics tools. Therefore, it could become a key data layer for AI healthcare applications and pharma R&D workflows. Second, Reddit’s value is not just advertising and community traffic, but continuously generated data on human conversations, sentiment, intent, consumption decisions, and product experiences. The OP emphasizes that Reddit’s scale is only a very small fraction of Meta’s, and argues that the market still compares Reddit using traditional social-media metrics such as users, advertising, and monetization efficiency, while not fully pricing in its “data oilfield” characteristics. The OP cites Reddit CEO Steve Huffman calling user content the “modern oil” of AI, as well as Tempus CEO Eric Lefkofsky’s positive comments on the long-term opportunity in the data business, to support the view that data licensing, API monetization, and AI partnerships will become core business lines. The image attachment is empty, so the main post contains no additional chart or screenshot evidence. On valuation, the OP does not provide a rigorous DCF, revenue multiple analysis, or option payoff calculation, and instead relies mainly on a relative narrative: infrastructure trades are crowded, while the data layer is still in the early stages of re-rating. Reddit’s market-cap gap versus Meta is used as upside imagination, while Tempus is framed as a beneficiary of scarce healthcare data. Potential catalysts include AI companies renegotiating or expanding data-licensing agreements, renewals of Reddit’s agreements with Google/OpenAI or litigation setting a pricing anchor, Tempus signing more data and modeling contracts with large pharmaceutical companies, and the market rotating from a compute-scarcity narrative to a high-quality proprietary-data scarcity narrative. The most important implicit assumptions are: frontier models will continue to rely materially on fresh, real, licensable human and professional-domain data; data owners will have sufficient bargaining power; and licensing revenue can translate into enterprise value in a high-margin, repeatable way. The main weaknesses are also clear: Reddit data contains substantial noise, bot content, and low-quality discussion, and the marginal value of incremental training data for improving large-model capabilities is disputed; the market-cap gap between Meta and Reddit is not only about data, but also about differences in user scale, ad infrastructure, and profitability; although Tempus has scarce healthcare data, medical compliance, data integration, sales cycles, and infrastructure investment may pressure margins; in addition, the OP’s option price targets are aggressive and lack quantitative validation of current valuation, revenue contribution mix, and competitive substitution paths.