We Made AI Analyze Everything We've Ever Said

In their 200th episode, the hosts review nearly 200 transcripts to surface recurring AI investment themes, track how the landscape has changed over 14 months, and outline where they believe the puck is going. They celebrate early correct calls such as ‘compute is king’ and the memory chip thesis, while highlighting emerging areas like space-based model training, local AI models on edge devices, specialized applications in drug discovery, and robotics at scale. The conversation remains broadly optimistic and frames SpaceX and Tesla as key beneficiaries of the next phase of AI deployment. - Retrospective analysis of 14 months of AI investing themes using LLMs on full transcript archive - Early calls proven correct: compute as national security, memory chip windfall (Micron, SK Hynix, Samsung up ~153%) - Notable private-market returns from podcast guests (Valor Atomics 13x, OpenRouter, Zipline, etc.) - Fading buzzwords: superintelligence, AGI, crypto; rising focus on IPOs, memory, and quantum - Shift in model-lab dominance: Anthropic overtakes OpenAI in show mentions, driven by Claude Code adoption - Space-based AI training emerges as a critical new frontier, with SpaceX positioned as the leader - Growing interest in local/edge models, personal AI agents, and new device form factors (Apple, OpenAI) - Bullish long-term view: hosts have a 2.8:1 bullish-to-bearish mention ratio across 200 episodes