Explains origin in shorting fraudulent Chinese reverse mergers after witnessing systemic dishonesty in microcaps.

His father's experience and the Potemkin factory visit to Orient Paper revealed a $150M market cap company with real revenue of only $2-5M.

Posits an inverse relationship between interest rates and dishonesty in society. Low rates anesthetize investors to risk, pushing bad behavior from microcaps into mid-caps.

The 'grey zone' of legal but economically misleading behavior is now perhaps the norm.

1) Easier shorts exist than Nvidia. 2) Flows, especially passive, drive parabolic price moves via float squeeze, creating a 'technical value' beyond fundamentals. 3) His view has '180'd' in the last month from sanguine to deeply concerned.

The right question is not about AI company valuations, but what AI will do to society and the market.

Points to private credit, citing opacity, lax documentation in ABS/CLOs, and allocator herd behavior reminiscent of the internet bubble.

Anecdotes of missing lien release filings and CLO managers not servicing/tracking underlying loans. Connects this to potential AI job loss intersecting with credit problems.

It is not unrealistic that AI displaces 15% of US knowledge work jobs in 3 years, possibly 20-25% in 5 years.

AI (Claude) dramatically improves internal communication and report drafting, saving weeks of work. It erodes his edge in writing but aids in accounting analysis.

It helps turn an analyst's patterns into clear memos and cites accounting standards (ASC).

The catalyst is not valuation but technicals. Cites Mike Green's thesis: when passive inflows from target-date funds taper and reverse to net outflows, a 1929-magnitude crash could occur.

AI-induced job losses could cause laid-off knowledge workers to stop 401(k) contributions and then make early withdrawals. This reversal of flows with no buyer to 'catch the falling knife' would test Green's thesis.

Five weeks ago he was completely sanguine; now his view has '180'd' on this risk.

The long side in hyperscalers is problematic as they are index components. Instead, they are creating 'highly convex trades,' essentially shorting credit where spreads are tight and vol is low.

Positioning for convexity with capped potential loss, as in the lead-up to the GFC but without a deep CDS market.