Hosts Debate AI Debt, Bond Market Signals, and Whether Stock Market Carnage Is Hidden Beneath the Surface

Michael Batnick pointed to two headlines from the same publication, the New York Times, about two weeks apart. Two weeks earlier, the paper said high interest rates were not slowing the AI boom, which he described as a problem for the Fed. The most recent headline said weak links were starting to show in AI borrowing.

Michael said: "High interest rates aren't slowing the AI boom. That's a problem for the Fed. Then last week: weak links start to show in AI borrowing." Ben Carlson called the shift hilarious and said it felt as if the story had flipped immediately to the claim that nothing the Fed could do would matter. The hosts treated the speed of that reversal as the point rather than any single data point.

The hosts said that in the prior three episodes the 10-year yield kept rising and the stock market appeared not to care, until a Tuesday when a hot inflation report arrived. Michael said he did not think the timing was random: investors had been thinking inflation was not going away and that the business cycle might be nearing its end, with fuel prices not the only cause.

Ironically, yields came down sharply after the hot inflation print. Michael recalled a similar episode in October 2022, when inflation was running hot and stocks were down about 20%. A bad inflation report then caused stocks to open down 2% before the market, and by the end of the day they were up 2%. He called that day the market bottom on bad news.

Michael predicted the same thing could happen in the bond market, with a bad inflation print leading people to believe the problem had been identified and bad news being fully priced into bonds. Ben suggested this could mean rates topping out while stocks bottom, noting that the breadth in the stock market had been poor even as the index hit highs and rates kept rising.

Ben asked whether the market might save itself from a blow-off top bubble again, because these companies now carry debt and the bond market is the ultimate judge of risk. Michael said he was glad Ben had made that point.

Ben said the bond market could put a ceiling on the speculation. If it is the thing that says no, this is good news, because purely private transactions in equity markets can go wild. Michael restated the idea as the cost of capital mattering: when risk assets are not priced for risk, as in a zero-interest-rate environment, behavior can become very sloppy, while a governor on investor appetite for risk lets markets function more healthily.

The hosts then argued about whether CoreWeave would still have had high bond yields in a zero-rate environment. Ben's answer was that zero rates are far worse in general than a normal cost of capital, though he drew a distinction between household balance sheets, which he said benefit from lower borrowing costs, and corporate balance sheets.

Ben said the railway bubble is the example everyone keeps returning to in discussions of AI, and that even with a useful infrastructure there was still a crash. He added that the railways of the 1800s probably went through three separate booms and busts rather than one.

He argued that people currently expect a single big boom followed by one big bust. In his view it is more likely that AI will see several booms and several busts, without one huge blow-off top that brings the whole system down, with the pattern being a boom, a mini bust, a mini boom and another mini bust.

Michael agreed and said there had already been three episodes of AI excitement: DeepSeek, situational awareness over the summer, and the current one. He added that investors' risk appetite cannot switch off immediately, since years of taking risk and being rewarded have built up strong habits.

Michael linked the AI trade and the amount of money companies were borrowing to a possible slowdown in borrowing as the economy weakens. He said his final answer was about a 15% chance that CoreWeave is a canary in the coal mine. His reasoning was that hyperscalers would not be affected, since CoreWeave's debt is trading at higher yields.

Ben disagreed, calling CoreWeave a small company and a drop in the ocean in AI. Michael noted that Ben had opened the show by calling a mortgage company that went under nothing. Ben replied that the size of a company does not determine whether it is the real signal. He said there would eventually be a canary but that this was not it.

The hosts also cited the week's performance: an equal-weight version of the Nasdaq 100 trailed the S&P equal-weight index by 4.5 percentage points, its largest weekly underperformance since November 2020. Consumer staples outperformed tech, with staples recording their best week in 24 years.

Ben argued that many stocks have had massive years and should have experienced drawdowns, because returns that high cannot happen without a 20% drop. Michael questioned whether the context was fair, noting that the stocks in a Bespoke list had all been up more than 150%, and that the average stock was up a decent amount this year.

Michael then described a college student's website called Portfolio Autopsy. It looked at the 343 companies in the S&P in 1999 that still exist today and found that 92% of today's S&P 500 companies had lost at least half their value at some point. The worst drawdown for the S&P this century was about negative 55 to 56%, while 85% of surviving companies fell further than that at some point. Examples included JP Morgan falling 74%, Microsoft's peak-to-trough drop of 70%, Cisco falling 90%, and Nvidia falling 90% at one point.

Michael cited Jim Cramer's tweet, which said billionaires are always bearish and to pay them no mind. Ben added that legendary investors who are white-haired are often bearish, and Michael said that being a hedge fund manager means being bearish almost all the time because you are a risk manager.

Michael also described an old 1960 paper showing the S&P 500 then comprised 425 industrial stocks, 50 utilities and 25 railroad stocks, and said that composition held until the 1980s. He then turned to Jason Zweig's Wall Street Journal piece on how the stock market has changed. Zweig noted that trading used to be far more expensive, with brokerage fees of up to 2% and mutual funds charging 5% to 8% upfront.

Ben said the playing field was tilted until Reg FD in 2000. He described how Jesse Livermore's tactics stopped working after the Securities and Exchange Commission was founded in 1934. Zweig argued lower future returns could follow from fewer risks, while the hosts countered that today's lower fees offset that on a net basis.

Ben said that if the largest companies had been growing earnings at 25% a year for 15 straight years, the returns would be good, and that trading costs barely matter by comparison. He said that, ten years earlier, anyone who knew what would happen with the hyperscalers would not care how much it cost to trade a stock.

Michael added that the real point is that investors must ride the waves when they happen rather than repeatedly trading in and out and getting out of the way.

Ben joked that Michael, a Midwesterner, was a confrontational coward, and said New Yorkers like confrontation and seek it out. Michael replied that he likes playful confrontation with friends and that they like to argue.

A chart circulated last week showing a survey of Anthropic employees asked what they expected nominal GDP growth to be in 2030. The median answer was 17%, which Michael found striking.

Michael said tech people can help create the future but cannot predict it, because they often do not understand the real world. He pointed to stories that AI agents could cost banks $500 billion by winning better savings rates, and noted that JP Morgan blocked outside AI agents while Citigroup and Bank of America restricted them from accessing accounts. He argued that tech forecasters often miss real-world barriers to entry and how things can be shut off, which he said is why their predictions have often been wrong.

The hosts discussed a Politico piece about older homeowners who do not want to pay property taxes. It quoted a 72-year-old Akron resident, Ray Warner, who said he was old, had paid off his home, and asked why he should pay property tax. Social media reacted strongly against what it called entitled baby boomers.

The piece said plans to substantially reduce or repeal property taxes had emerged in more than two dozen states in the last two years. Georgia, Indiana and Wyoming had already cut residents' tax bills by billions of dollars, and governors in Florida and Texas competed to lower property taxes the most. In Ohio, the proposal did not pass because the governor argued that eliminating these taxes would force sales taxes up to about 17%.

Ben said that cutting property taxes shifts the burden onto people who do not own homes, who are less wealthy, and that they end up paying more in taxes. Michael said that even if older residents no longer have children in school, roads, firefighters and police still exist.

A tweet by Shil Manote noted that United now requires $32,000 of spend to earn $1,000 in rewards, compared with $10,000 to $15,000 before. The hosts said Disney is doing something similar, and Ben argued companies are waking up to the fact that wealthy customers can spend without limit.

Ben also described crowded premium airport lounges as evidence that rich customers are receiving generous benefits, and said such benefits should be cut. He then mentioned a $250 IV drip he had at a spa, which Michael questioned.

Ben said Netflix had taken a 20% loss, which he called a teachable moment about how the stock market works. The financials are strong, with the company earning billions of dollars in profit each quarter and healthy margins. Growth is slower than before but still positive.

Michael said the company is not growing nearly fast enough and that YouTube is eating its lunch. Netflix laid off about 1,000 people last week, and the stock barely reacted. Matt Belloni then reported that the head of content, Bela Bajaria, who oversaw a roughly $20 billion content budget, had left. A chart Belloni shared showed that up to about 30% of senior Hollywood executives were gone within two years of appointment.

Michael noted that over this decade the S&P 500 had outperformed Netflix by about 50%.

Netflix was the first platform to publish top-10 lists of what viewers were watching, and other platforms have since copied the practice. Michael said Peacock's top-10 was especially poor and that some lists on other platforms made him wonder what was wrong with the people watching.

Ben said Ari Emanuel, speaking with Ben Thompson, discussed how AI will change Hollywood, and said that is one of the things technology may take from people. Ben also described watching 90s action films with his son, including Cliffhanger, Terminator 2 and Speed.

Michael said he saw Hokum, starring Adam Scott and directed by Damian McCarthy, who had made Oddity, a horror film he enjoyed. He said he needed a podcast to explain what happened in Hokum, which was not very complicated. He also described watching a horror movie with his son, Five Nights at Freddy's.

Ben recommended Tony, a fictionalized coming-of-age film about Anthony Bourdain, rating it 7.5 and saying he loved it. He said it draws on Kitchen Confidential and features three young actors from The Holdovers, Task and The White Lotus, along with Antonio Banderas. He also discussed Ben Mezrich's The Accidental Billionaires, which he described as about 60% true and 40% sensationalized, and which relied heavily on the Winklevoss twins. Ben said he still could not believe a $2 trillion company was started like that.

Michael said his favorite story from a Keith McNally book was about a $2,500 bottle of wine being mixed up with a $10 bottle, and the diners at the table did not realize it. Duncan was jokingly named as owning the craziest AI stocks in a poll asking whether the market has carnage under the surface or nothing to worry about.