By Thierry Hasse, Chief Investment Officer
Elevage Partners | August 11, 2026
A Labor Market Shrinking, Not Healing
The Bureau of Labor Statistics delivered a jobs report Friday that looked worse than it was, and better than it was, in the same release. Nonfarm payrolls fell by 23,000 in July, missing a consensus estimate near 83,000 (Source: Bureau of Labor Statistics). The bureau also cut May and June by a combined 103,000, with June’s initial gain of 57,000 revised down to just 20,000 (Source: Bureau of Labor Statistics). The unemployment rate, meanwhile, slipped to 4.1% from 4.2%. In our assessment, that decline is the least reassuring number in the release.
The practical consequence, in our view, is a labor market that can look stable on the headline rate while growing structurally weaker underneath it. A shrinking workforce caps how fast the economy can grow without inflation, tightens the math on Social Security and pension funding, and pushes employers who lean on younger workers, retail among them, toward higher wages or faster automation. It also muddies the Fed’s read: a falling unemployment rate argues against cuts, a falling participation rate argues the opposite. Markets chose the second interpretation Friday, and Treasury yields moved lower on the release (Source: CNBC).
When AI Enthusiasm Meets a Margin Call
The week’s other lesson arrived from a hedge fund rather than government statistics. Leopold Aschenbrenner, the 24-year-old former OpenAI researcher, built Situational Awareness into a $45 billion fund on concentrated bets in AI infrastructure names including SK Hynix, Nebius, and CoreWeave. In July, those positions reversed, and the fund lost roughly 67% of its value (Source: CNBC). The fund carried leverage of roughly four times capital, so a 25% decline in the underlying stocks was enough to erase its equity (Source: Barchart).
When margin calls arrived from Goldman Sachs, JPMorgan, and Bank of America, Aschenbrenner had no room left to maneuver, and by July 30 the fund had sold its entire public equity book in one motion (Source: CNBC). Citadel, a firm with a documented history of buying distressed books from blown-up funds, stepped in and bought roughly $16 billion of the portfolio at a discount of about 10% (Source: The Wall Street Journal).
It is the same story every cycle tells: concentration plus leverage removes an investor’s ability to control downside risk, and this time the prime brokers moved faster than they did during the 2021 Archegos unwind. What is remarkable is not that leverage turned a concentrated position into a disaster. That lesson repeats every cycle. It is that the fund was making, in extreme and borrowed form, the same wager the broad market is making right now, that AI infrastructure is the trade that keeps paying. The unwind of a $45 billion fund barely rippled beyond Wall Street, not because investors missed it, but because the market is not frightened of that bet. It is committed to it. Concentration and leverage, not conviction, decide who survives when the bet turns.
Priced for AI to Deliver
None of this stopped equities from doing what they have done most of this year, which is making new highs. The S&P 500 closed Friday at 7,757.64, its 26th record close of 2026, capping a 3.6% weekly advance (Source: CNBC). The Nasdaq gained 5.2% on the week, with chip stocks leading the bounce (Source: CNBC).
Underneath the rally sits a genuinely strong earnings season. As of Friday, 88% of S&P 500 companies had reported second-quarter results, 86% of them ahead of estimates, above the five-year average of 78% (Source: FactSet). Excluding one-time investment gains booked by a handful of companies this quarter, earnings grew 32% over the prior year, the best pace since 2021 (Source: FactSet). The scale of the improvement shows up just as clearly in dollar terms: the index earned roughly $273 per share in all of 2025, and analysts now put the calendar 2026 estimate at $351, an increase of nearly 29% that has been revised higher every month this year (Source: FactSet; First Trust Advisors).
That is a lot of good news, and the market has priced in a great deal more of it. Growth is expected to moderate to 13.6% in 2027, with the forward price-to-earnings ratio, the price investors pay for each dollar of expected earnings, at 20.0, above both its five- and ten-year averages (Source: FactSet). Twenty times earnings is not, on its own, alarming. What turns it into a bet, in our assessment, is the assumption underneath it, that the AI investment cycle keeps converting into profits at close to this pace. Strip that assumption out and the multiple looks expensive. Leave it in and it looks fair. The entire question is which one proves right.
For now, the market expects the two questions hanging over it, the Fed’s path and the war in Iran, to resolve in its favor. Washington and Tehran are reportedly closing in on a deal to reopen the Strait of Hormuz (Source: Axios), and attention turns this week to inflation data and whether a softening labor market gives the Fed room to cut (Source: CNBC).
It has been a disorienting week to be an investor: a jobs report that reads as both weak and reassuring, a $45 billion fund unwound almost without notice, and a market at record highs through all of it. Step back, and the three are one story. The record highs are not built on blind optimism. They rest on a single bet, that the money pouring into AI keeps converting into real earnings, and the strong quarter just reported says that bet is being paid so far. But a market leaning this hard on one question turns unforgiving, rewarding the companies that prove it and punishing the ones that miss. A fund that made the same bet with borrowed money simply vanished.
What matters is not which way the AI question resolves. We do not know, and neither does the market, which is why it is still waiting to be shown. The point is that the largest risk in this market is no longer overpaying for a good company. It is holding a portfolio that quietly depends on a single outcome. Valuation discipline guards against the first. Owning genuinely different kinds of businesses, with their fortunes tied to different questions, guards against the second. In our view, that is what lets clients stay invested through weeks built to punish anyone who bet everything on being right.