Antonio Linares runs a concentrated six-stock portfolio plus micro investments, focusing on 'singularity scalers' — companies with proprietary data moats that compound value via AI scaling laws. He bought Palantir at ~$7, held through 2022-2024 drawdowns, and teaches a course on modern value creation processes. His framework emphasizes betting early on extraordinary companies, holding forever, and learning from concentrated mistakes.
AMD's decade-long investment in chiplet architecture and Xilinx FPGA technology uniquely positions it to capture the emerging physical AI/edge AI market, with data center growth already inflecting and embedded segment turning around.
Duolingo's proprietary behavioral data and ontology velocity create a compounding AI advantage that drives accelerating user engagement, content shipping, and free cash flow per share, while the market incorrectly prices in AI disruption risk.
Tempus AI has built an irreplicable proprietary dataset (500 petabytes, 45M+ patients, 5000 institutions) by embedding diagnostic pipelines into US hospitals, creating a Palantir-like flywheel where generative AI turns expanding multimodal data into exponentially more valuable clinical insights at near-zero marginal cost, evidenced by 126% net revenue retention and $2B+ in pharma data licensing deals.