Leopold Aschenbrenner, the former OpenAI researcher turned investor, flipped bearish on NVIDIA, AMD, and Broadcom in his Q1 13F filing. The move signals a thesis shift: the semiconductor trade is crowded, and the next bottleneck in AI infrastructure is no longer chips.
The video argues the real opportunity has rotated to power, data centers, and energy infrastructure. CoreWeave gets specific attention as a proxy for where capital is actually flowing. The hosts break down why retail is still chasing semis while institutional money is repositioning stack-wide.
The tension worth reading into is the 'where the trade breaks' section at the 16:31 mark. Aschenbrenner built his reputation on predicting AI scaling trajectories correctly. If he is wrong about the semiconductor pullback, the reasoning behind the error matters as much as the error itself.
[WATCH ON YOUTUBE →]