What Leopold Actually Got Wrong

Hosts Josh and Ejaaz recount the rapid implosion of Leopold Aschenbrenner's hyper-leveraged AI infrastructure hedge fund after positions were hunted and liquidated by Citadel's Ken Griffin. They examine timeline, predatory selling, and the remaining Anthropic stake, then debate whether Leopold was fundamentally wrong. They conclude the AI infrastructure demand thesis remains intact and the liquidation-driven sell-off created a long opportunity in names like SK Hynix and Bloom Energy. - Leopold Aschenbrenner’s fund ran from ~$225M to $4.5B on 4x leveraged AI infrastructure bets before collapsing in weeks. - Market participants hunted his known positions, triggering margin calls and a forced block sale to Citadel at a steep discount. - Ken Griffin / Citadel repeated a pattern seen in Enron and Amaranth, profiting billions from the distress. - The public equity book was entirely sold; the surviving value is an early Anthropic stake that cushioned the drawdown. - Immediately after liquidation, the same AI infrastructure names rallied 20%+, suggesting fundamentals were never broken. - Hosts maintain that mass hyperscaler capex into memory, chips, and power infrastructure keeps the AI buildout trade compelling. - The episode warns against extreme leverage and notes software/cyber outperformance as a potential counter-thesis.