SquareBook

Market Colour

Week of 1–7 August 2026 37 sources tracked — 20 core + 17 bench 5 conversations, 16 singles a 4-minute scan Past editions →
The big conversations
Funds9 of 37 sources on this · new this week

Anatomy of a blow-up: the Situational Awareness post-mortems

Leopold Aschenbrenner’s AI fund was caught in late July’s margin-call cascade after two years of huge gains — and this week everyone from Marc Rubinstein to the podcast tier wrote the autopsy. Accounts differ on how terminal it is.

Where they agree

  • Leverage plus concentration did the damage — the fund’s thesis wasn’t “wrong” so much as margin-called before it could be right.
  • It’s a systemic warning, not a one-off: All-In framed it as a $20bn margin call inside a crowded trade everyone owns.

Where they split

  • Net Interest reaches for Amaranth 2006 — a stress event that didn’t end the strategy; Prof G says the collapse “is just the beginning.”
  • The tape settled the argument mid-week: Aschenbrenner reportedly put $400m into a private company on Tuesday, days after the margin calls — vindicating TBPN’s “Leopold is Back!” and, as Levine notes, keeping the thesis while dropping the leverage that broke it.
AI & Markets9 of 37 sources on this

The whipsaw week: record highs on Monday, top-calling by Friday

Late July’s AI selloff reversed violently — the S&P closed above 7,700 for the first time on Monday — then wobbled again on oil and rate-hike odds. The bulls point to demand; the bears now include Michael Burry.

Where they agree

  • The violence of the round trip — correction to record high in three sessions — is itself the story: leverage, not fundamentals, is setting prices week to week.

Where they split

  • Gavin Baker (on ILTB) sees no slowdown in real AI demand; Burry (via Prof G) is calling the top outright.
  • The Bear Cave flips the frame: forget the AI winners — fifty companies are being quietly eaten by it.
  • Stoller asks the auditor’s question: are the big AI firms’ books even right?
Where they sit
MB IL PR OV MD BC PG
◂ Buy the dipThis is the top ▸
Solid rings = core sources, dashed = bench. Positions are our editorial judgement, not the authors’ own words.
IPOs4 of 37 sources on this · thread returns

The SpaceX lockup expired — and a thousand SPVs met their moment of truth

The stock held up; the plumbing didn’t. Thursday’s first release was also the day a lot of retail investors found out whether the pre-IPO vehicle that sold them “SpaceX exposure” ever owned the shares at all. Bankers reportedly count at least a thousand SPVs holding SpaceX stock.

Where they agree

  • The feared flood of unlocked stock didn’t sink the price — the damage landed on holders of synthetic and SPV-wrapped exposure instead.
  • A transfer restriction is a wonderful hiding place: while nobody could take delivery, nobody could check. The lockup release is the audit.

Where they split

  • Levine inverts the usual moral — it isn’t the crash that exposes the fraud, it’s the boom: a bust gives a fake fund an alibi, while a triumphant IPO makes investors ask for their winnings.
  • TBPN frames it bluntly as investors getting “rugged”; Merryn’s emergency episode plays it as behaviour: FOMO is not an investment strategy.
Fraud & Governance4 of 37 sources on this · second week

The “golden age of fraud” thread hardens

Last week it was a vibe; this week it has names and mechanisms — from Chanos’s late-cycle warning to a founder who ran her “AI” by hand.

Where they agree

  • Speculative peaks and weak enforcement breed grift — and the AI boom is supplying both conditions at once.

Where they split

  • Chanos reads fraud as a cycle signal; Stoller as structural; TWiST tells the fake-it-till-you-make-it story almost admiringly — the line between grift and hustle is half the debate.
UK Markets5 of 37 sources on this · thread reheats

“Britain urgently needs to fix its stock market” — the LSE thread is back

Two weeks after the de-listing gloom, the conversation turned constructive: cheap UK trusts, what Schroders’ new CEO does about it, and whether the Burnham government gets any of this.

Where they agree

  • UK equities are cheap enough that the discount itself is now the opportunity — and fixing the market is squarely a government problem.

Where they split

  • Money Distilled wants policy urgency; Undervalued Shares says just buy the trusts at a discount while the politicians argue.
The thread board — how the conversations moved
AI bubble debate▲ whipsawedCorrection → record high → top-calls, all in five sessions. Now inseparable from the leverage story.
Fund blow-ups● new threadSituational Awareness post-mortems everywhere; watch for who else was in the crowded trade.
SpaceX post-IPO▲ returnedLockup expired 6 Aug, stock up ~6%; the SPV-discount saga is the tail to follow.
Golden age of fraud▲ hardeningSecond consecutive week, now with specific names and an AI-accounting angle.
UK: shrinking LSE▲ reheatedBack, and constructive this time — trusts, Schroders, policy fixes.
Open vs closed AI▼ simmeringA few hits (20VC on a $100bn open-source model; OpenAI–Hugging Face fallout on TBPN) but no longer the lead.
Warsh Fed● forming~65% September-hike odds, a US–Japan yen intervention (Odd Lots), gold stirring. Candidate cluster for next week.
Stripe / PayPal▼ silent 2 wksMoving to the watch list; next catalyst is regulatory.
Private credit▼ simmeringOne direct hit (MiB with Future Standard’s CIO, 5 Aug); leverage themes are adjacent everywhere.
Disagreement of the week
Worth your time — three picks
1

Leopold’s Fall

Situational Awareness and Amaranth, twenty years apart — the definitive write-up of the week’s defining story, from the one writer who does fund post-mortems properly.

2

AI, Productivity, and Rates: Part I

If AI really raises productivity, rates may need to be higher, not lower — the least-discussed and most consequential angle on the whole boom. A bench pick earning its place.

3

Tick sizes, or why everyone suddenly loves prediction markets

The cleanest market-structure argument of the week: an event contract is worth about 50 cents and trades in penny ticks, so the spread is ~2% of notional — roughly 100× the equivalent on a share of stock. Robinhood’s quarter bears it out: a fraction of the volume, more of the revenue. If you care how venue design quietly decides who captures the economics, read this one.

The rest of the tape

Deals & companies

Markets & macro

The craft of investing