THE WEEKEND SPREAD

This Week: Meta's quarter made the AI argument, and the build underneath it is a third bigger than the printed number. France shows what happens when a government runs out of arithmetic. Brazil votes Sunday in a tie. And a series on the great financial crises starts, the overview and 1929, so we can revisit the mistakes and not make them again. Hope you enjoy this issue.

Subscribe: Forward this to a friend, or subscribe at theweekendspread.com.

Key Takeaways

1. Meta's quarter and the AI build are the same story. The quarter was strong. The build underneath it is the story: roughly a third bigger than the printed number, about $3.03 trillion of commitments, most of it off balance sheet. Every dollar of that build is revenue to someone. (AI Paranoia · Six Times Bigger, While Growth Is Cheaper)
2. France is the fiscal story to watch. The deficit, the debt and the borrowing cost all point the same way, and the street has noticed. (The Street and The Deficit)
3. Brazil votes Sunday, positioned one way and hedged the other. The polls call the runoff a tie, but first-round money poured in as if Bolsonaro wins, and the options market is paying double for protection against the opposite result. The piece inside shows the tape, and how I am hedged. (Positioned One Way, Hedged the Other)

🇫🇷This Week's Highlight · The Tape

The Street and The Deficit

France had one of its most violent days of protest in decades on Saturday, and behind the street is the arithmetic: a deficit of 5.1% of GDP in 2025, debt at 119% of GDP, a postwar record, and a 10-year borrowing cost through 5% this week, the highest since 2002. On Friday, Moody's kept France at Aa3 with a negative outlook, sparing the government the downgrade the market expected. For the first time in decades, the French state cannot simply pay the street to go home.

My take: France's problem is no longer that the street can stop a reform. Every previous French crisis ended with a check. This one cannot.

Featured Reads

The Week in Numbers

The Tape

Six Times Bigger, While Growth Is Cheaper

Twenty chip and AI hardware companies were worth $2.39 trillion the day ChatGPT launched. At Friday's close they were worth $15.38 trillion, 6.4 times more. That is the bubble chart. The other chart does not fit it: over the last twelve months the price of a point of earnings growth has been falling, and on the S&P's tech names it has halved. Value up, the price of growth down, at the same time.

The Tape

AI Paranoia

Meta reported a strong quarter on Wednesday night and the stock still fell 7% on Thursday. Tesla beat and fell 5%. Meta was sold because it raised its spending plan. Tesla was sold because it cut its spending plan. When both directions of the same decision produce the same trade, the decision is not the information. The mood is.

The Lab

The Financial Crisis Series Launches

This week The Lab starts a new series: the great financial crises, taken apart one at a time, in the order the world lived them. The point is simple: revisit history, understand the mistakes that were made, and do not make the same ones again. It opens with a short history of the pattern, then goes straight to the crash everyone still measures the rest against, 1929. One a week from here: Brazil, Black Monday 1987, 1998, the dot com bubble, 2008, Covid, and finally the question of whether the AI boom belongs on the list.

A Brief History of Financial Crises

Financial crises are older than banks, older than stock markets, older than paper money. The first one we have a record of hit Rome in the year 33. Land prices fell, credit froze, and lenders called in their loans all at once. The emperor Tiberius had to inject 100 million sesterces into the system to stop the spiral. Two thousand years later, central banks do the same thing with bigger numbers. The machinery changes. The crisis does not.

Here is the short history, and the pattern hiding inside it.

1929: The Crash, and the Depression That Did Not Have to Follow

Every crisis in this series gets compared to 1929. There is a reason. 1929 is not one crisis. It is two crises stacked on top of each other, and only the first one was unavoidable.

The first crisis was the crash: a speculative bubble built on borrowed money, and it burst the way bubbles always burst. The second crisis was the Great Depression: three and a half years of bank runs, deflation, and mass unemployment that turned a bad recession into the worst economic disaster in American history. The crash lit the fire. Policy poured the fuel. A stock market crash did not have to become a depression. This one did, because the people in charge made the wrong choice at almost every turn.

Chart of the Week

The average 30 year fixed mortgage rate spent nine weeks climbing, from 6.63% in early August to 7.40%, as the 10 year Treasury climbed through 5%. This week it fell, to 7.32%, the first decline in nine weeks. The reason is the week's strangest trade: investors sold AI stocks and bought Treasury bonds for safety on Thursday, pushing the 10 year yield back below 5% even as oil rose above $112 a barrel. Money running from risk found the bond market first, and the mortgage rate followed it down.

My take: one down week is not a trend, and 7.32% is still a heavy rate. But watch the mechanism. If the AI selloff keeps feeding a bid for bonds, the rate American households pay falls while the stock market does the worrying.

Freddie Mac's release: Mortgage Rates Decrease to 7.32% (Morningstar) →

The Study Behind the Headline

A Busy Social Life Tracks With a Sharper Brain

The headline version: staying social keeps your brain young. The study, pooled from five long running surveys covering 57,793 adults in 24 countries, followed for a median of six years, is more specific. What tracked with sharper thinking was variety, not just company. People who took part in more kinds of social activity, volunteering, sports clubs, classes, kept better cognitive scores as they aged, and those doing three or more kinds improved the most year to year. Volunteering and sports clubs stood out in every survey. The study shows a link, not proof of cause. The pattern worth copying is plain: the people who kept doing different things with different people kept their edge.

My take: the variable was variety. Same friends, same table, same conversation does not count. Mix the inputs and the brain keeps working.

Source: mindbodygreen, Oct 22, reporting a study in the Journal of Epidemiology and Community Health. Reading time: about 4 min.

Quote of the Week

"Poor countries are poor because those who have power make choices that create poverty. They get it wrong not by mistake or ignorance but on purpose."

From Why Nations Fail, this issue's What I'm Reading. Poverty, on this account, is rarely a failure of knowledge. It is a set of choices, kept in place by the people the choices benefit. That is why good advice alone never fixes it.

The book: Why Nations Fail, Penguin Random House →

Culture

This week in Culture: a song, a show, a book, and a newspaper. What I'm listening to, watching, reading, and reading in print. See it all on the site →

Listening: "Take Five," The Dave Brubeck Quartet (1959). Paul Desmond wrote it in 5/4, a meter that should not swing and does. Three minutes, and the room resets. Listen →

Watching: TYSON, Netflix (2026). A four-part documentary on the life of Mike Tyson, directed by Floyd Russ. Brooklyn to the youngest heavyweight champion in history, then the fall, and the long way back. Now streaming. Watch →

Reading: Why Nations Fail, Daron Acemoglu and James A. Robinson. The 2012 book that won its authors the 2024 Nobel Prize in economics, shared with Simon Johnson. Why are some countries rich and others poor? Their answer is institutions: the ones that spread power and opportunity, against the ones built to extract both. Robinson, the co-author, is also this week's podcast guest, applying the same test to the United States. The book →

In Print: The Print Edition, David Coggins and Michael Williams. A quarterly newspaper, on paper, about travel, style and sports. An escape from the algorithm, from people who still believe in the newsstand. The paper →

Podcast Picks

Most Read

• The Lab: "Inflation, Explained," what it is, where it comes from, and why a dollar from 1913 buys what about 3 cents buys today.
• The Tape: "Nobody's Getting Fired. Nobody's Getting Hired." A labor market frozen solid, and why the freeze matters.
• Travel: "Two Weeks in Portugal," the itinerary Sachenka and I keep going back to.

The Week Ahead

The Fed and the biggest earnings week of the season land on the same day.

• Brazil's presidential runoff is Sunday, October 25. The Brazil piece in this issue is the setup.
• The Federal Open Market Committee meets Tuesday and Wednesday, October 27–28, with the rate decision on Wednesday at 2:00 PM ET.
• US durable goods orders are released Tuesday, October 27.
• Microsoft and Alphabet report earnings Wednesday, October 28, the same evening the Fed decides.
• Thursday, October 29 brings the advance estimate of third-quarter GDP, the PCE price index, and the weekly jobless claims.
• Amazon reports earnings Thursday, October 29, after the close.
• Early voting winds down: Georgia's in-person early voting runs through Friday, October 30, ahead of Election Day on Tuesday, November 3.

The Fed's meeting calendar: federalreserve.gov →

The Week in the Markets

Friday, October 23 close.

Indices, FX and commodities in percent; Treasury yields in basis points.