Limitless podcast hit episode 200 and did something useful: reviewed nearly 200 transcripts to audit their own calls. The two recurring investment themes that held up across 14 months were compute and memory. The Cursor thesis paid off. Early bets on chip infrastructure were validated. This is not a victory lap disguised as analysis, it is an actual retrospective with named companies and tracked returns from their portfolio.
The episode maps which trends are rising and which are fading. Google's AI comeback gets its own timestamp. So do local models, recursive AI, geopolitics, and space-based model training. That last one is not a throwaway: specialized AI applications and on-device inference are framed as the next structural shift, not a sideshow. The hosts are explicit that turning intelligence into value, not just building it, is where the next 200 episodes point.
Read the full episode if you want the specific companies they revisited and the logic behind why compute and memory remain the load-bearing themes even as the application layer gets crowded. The geopolitics and recursive AI section near the end is the part most publications are not covering seriously yet.
[WATCH ON YOUTUBE →]