To get John Authers’ newsletter delivered directly to your inbox, sign up here.

Today’s Points:

• The US credit market is hunting for cockroaches.
• The future of AI is agentic; it won’t need you to ask first.
• S&P downgraded France just as it looked like the PM had found a way to survive.
• China’s growth weakens — but it still has rare earths.
• AND: Some great movie jewel heists, from Clouseau to Wallace & Gromit. 

Cockroaches

Words matter, particularly when data-dependent markets are starved of new information by the US government shutdown. That’s been rammed home twice in the last week, as the US sold off sharply first in response to trade threats by President Donald Trump on social media, and then to a comment from JPMorgan’s chief executive, Jamie Dimon, likening problems with auto-loan credit to “cockroaches.” As he put it, “When you see one cockroach, there are probably more.”

The latest trade spat, sparked by new controls on exports of Chinese rare earths, remains unresolved — and it’s hard to see any resolution for a while. China’s dominance of the category remains by far its strongest point of leverage over the US.

The credit issue is different. Dimon was responding to two bankruptcies for auto lenders, and his comments set off an intra-mural Wall Street squabble over whether the blame should rest with banks or with private credit lenders. That’s not relevant to asset allocators, who want to know whether a broad credit crisis lies ahead. Instead, investors seem to have sold in response to Dimon’s provocative phrase. The last delinquency data for auto loans, up to the end of the second quarter, already showed that there was plenty to worry about, with the worst credit quality since the Global Financial Crisis:

And yet it was not until Wall Street’s most famous man had spoken that high-yield bond spreads rose. Compared to the last major credit scares, surrounding the collapse of Silicon Valley Bank in 2023 and the Liberation Day tariffs this April, this still ranks as a minor event:

While volatility in stocks leapt last week, bonds were much calmer. Plentiful liquidity has kept turbulence low, which has been a necessary condition for the risk-on rally in other markets. Bonds continued to help out, with the 10-year Treasury yield even dropping below 4% last Thursday in a move that supports stocks:

The underlying issues remain clear enough. The US economy is locked into a “K-shaped recovery” in which some do well and many do not. This perversely keeps rates under control and eggs on the stock market — while allowing risks of social unrest to increase. For a clear K-formation, this is how the S&P 1500’s indexes of investment banks (buoyed by great trading results last week) and regional banks (hit by credit worries) have performed under Trump 2.0:

Last week’s selloffs reflect what Macquarie’s Viktor Shvets calls “the perception that underneath strong economic numbers, there are crevasses of credit and valuation risks that are deepening and broadening.” This is a reasonable fear, which extends from the very unbalanced economic recovery. Without good data, however, it remains a matter of conjecture — and that is why the cockroach comment had such an impact. The reaction showed enduring nervousness, but without harder evidence — meaning finding some cockroaches in broader economic data — the odds are that investors will  treat this as an opportunity to buy. 

AI: Size Isn’t Everything 

In the race to turn artificial intelligence from promise into profit, a new word has added itself to the vocabulary: “agentic” AI. What does it mean? ChatGPT answers the question. The key agentic traits, which ChatGPT seems happy to admit will render it obsolete, are:

1. Goal-directed behavior – pursuing objectives rather than individual tasks.

2. Autonomy – ability to act without continuous human input.

3. Planning and reasoning – ability to form and execute multi-step plans.

4. Environment interaction – taking actions that affect the digital or physical world (e.g., sending emails, trading stocks, controlling robots).

While the profit margins currently being generated by AI-fueled stocks still seem to justify their lofty valuations, fundamental questions remain about whether these massive investments will truly lift productivity and, in turn, sustain those margins over time. Three years after ChatGPT sparked the generative AI frenzy, the magic is still more promise than payoff, a reality that validates investors’ caution toward whatever new twist comes next. 

Still, that skepticism hasn’t dimmed the search for AI’s true use cases, which can sustain the momentum after the initial euphoria fades. Agentic AI, some hope, could be the true use. An AI personal assistant might not only schedule meetings but automatically negotiate times, book venues, and send follow-ups, all without any prompting. It’s the kind of AI that workers dread and executives dream of — it threatens jobs while lifting productivity. 

A plethora of such autonomous AI agents are at various stages of development. Notably, Walmart’s Sparky, which is currently an AI-powered shopping assistant that merely responds to shoppers’ queries, is on course to start taking autonomous decisions on customers’ behalf, according to analyst Robert Ohmes of Bank of America Securities. It will not rely just on successive prompts.

If this works as hoped, AI will identify what consumers want, weigh choices, strike deals, and finalize transactions — all while acting autonomously through a chain of reasoning steps that mirrors human intent. To quote McKinsey & Co.’s Katharina Schumacher and Roger Roberts, “This isn’t just an evolution of e-commerce. It’s a rethinking of shopping itself in which the boundaries between platforms, services, and experiences give way to an integrated intent-driven flow.”

McKinsey estimates that by 2030, the US business-to-consumer retail market alone could see up to $1 trillion in orchestrated revenue from agentic commerce, with global projections reaching as high as $3 trillion to $5 trillion.

That makes this sound like the “killer app” that has so far eluded impatient investors. The new SoFi Agentic AI exchange-traded fund, tracking companies that generate at least 30% of their revenue from agentic AI, offers a glimpse of how investors are positioning themselves to capture the next wave of the technology:

As the ETF, which tracks companies such as Tesla Inc., Palantir Technologies Inc., and Salesforce Inc., was launched last month, there’s little yet to glean from its performance. To some extent, this could be an exercise in betting on the “not-so-obvious” players — learning the lesson of the dot-com era, when many of the biggest names of the time eventually faded into obscurity. 

In June, a BofA survey found that 64% of organizations expect to pursue agentic AI initiatives this year. Most were only in the very early stages of deployment, with 53% in the exploratory and 25% at the pilot phases, and only 6% in production as of early 2025. Customer service, marketing, sales, and software development would be the first significant job functions to adopt agentic technology:

Doubts remain. In BofA’s most recent Agentic AI Handbook, Brad Sills  explains that investors have learned to be skeptical due to the relatively underwhelming adoption and monetization of generative AI. That leaves them cautious about forecasts of agentic revenue. Still, he argues that AI agents could be the catalyst that allows companies to make money out of AI by unlocking “sustainable, measurable, and material workforce productivity improvements.” 

Sills’ view echoes recent Nvidia research, which argues that small language models — rather than the large language models like ChatGPT that set off the AI frenzy three years ago — could form the backbone of the next generation of autonomous systems.

That’s mainly because the massive investments in computing capacity and power generation over the last three years have increased the attraction of smaller models. They are more energy-efficient, compact, and easier to deploy, and require far less computational power. They need less capex, and companies don’t have to fork out quite so much money to buy chips from Nvidia. The old belief that “bigger is better” is giving way to a new logic: Sometimes, smaller is smarter. Both model types will have their place, but it’s becoming increasingly clear that an all-knowing model is overkill for tasks that demand narrow, specialized expertise — particularly if they can be entrusted to systems that don’t even need to be asked first.

— Richard Abbey

Calling Clouseau

Just when things seemed to be improving in France, someone breaks into the Louvre and gets away with a priceless haul of jewels in an episode that seems to come straight from the Pink Panther movies. But it was possibly only the second-most damaging French financial heist of the weekend. Late Friday, Standard &Poor’s announced that it was downgrading the country’s sovereign debt. 

As recently as 2011, France was a AAA-rated credit, ranked very slightly higher than US Treasuries. Since then, all the three main rating agencies have downgraded it steadily, first thanks to the euro zone’s sovereign crisis and more recently because of the home-grown political crisis that broke out when President Emmanuel Macron called snap legislative elections in the summer of 2024. Bloomberg colleague William Horobin maps what has happened to the French rating over that time:

Once he’s finished catching the burglars at the Louvre, it’s tempting to suggest that Inspector Clouseau might be called in to help the similarly hapless Premier Sebastien Lecornu, who survived two confidence votes during the week. He did this by conceding to the Socialist Party’s demand that he shelve President Emmanuel Macron’s flagship pension reform, which would raise the retirement age to 64 from 62, until after the 2027 presidential election. That had an immediate effect on French bonds, with both the yields of 10-year OATS and their spread over equivalent German bunds dipping sharply. 

The relief would have been short-lived even without S&P’s intervention, as this fix deepens France’s fiscal problems even as it gets Lecornu out of the immediate political difficulty. Reducing spending on pensions would have eased the pressure on the budget, and Lecornu said that parties would need to agree on other measures to keep the government deficit below 5% of gross domestic product by 2026. It’s difficult to see how this can be done without resorting to measures almost as unpopular as raising the retirement age.

Even if he survives, economic uncertainty will persist. Standard & Poor’s said it expected French debt to reach 121% of GDP by the end of 2028 (from 112% at the end of 2024). “We expect policy uncertainty will affect the French economy by dragging on investment activity and private consumption, and therefore on economic growth,” with real GDP rising only 0.7% this year.

Not fully underwater, yet. Source: Michael Ochs Archives/Getty

Oxford Economics argues that by 2035, “We expect France’s public debt to reach 125% of GDP,” and warns that there is “still a risk that the government could be toppled before year-end if budget negotiations turn sour, which would likely trigger snap elections.” Barclays’ Saadalla Nadra-Yazji and Silvia Ardagna argued that Lecornu’s reprieve would be temporary, and that passing a budget by year-end remained “challenging.” 

If there are opportunities in the French situation, they are in equities. France’s stock market is dominated by multinationals, particularly its world-leading luxury goods manufacturers. The domestic economy has minimal impact on them, and yet the market’s stark underperformance of the rest of Europe since last year’s elections shows that investors are applying some kind of political risk discount to them:

The luxury goods sector enjoyed surprisingly strong results last week (another example of the K-shaped recovery), causing the entire French stock market to rally. Chinese consumers are still avidly buying their products. If there is an opportunity in the horrible French morass, it may lie in its stock market.  

Survival Tips

If Clouseau needs a hand... Photographer: MJ Kim/Getty

Returning to the Pink Panther movies and Clouseau. Peter Sellers was a genius. I was introduced to them via the spin-off cartoons, which were fantastic. It also had what was surely the greatest movie theme of all time, care of Henry Mancini. 

The only movie that comes closer to what happened this weekend at the Louvre is The Wrong Trousers, the Wallace & Gromit classic. But not even Wallace could help out the French budget at this point. Have a great week everyone. 

More From Bloomberg Opinion

• Hal Brands: China Is Already Winning the Trade War America Wanted
• Jonathan Levin: Bessent Is Reverting to His Hedge Fund Ways With Argentina
• Matthew Yglesias: What Makes This Shutdown So Different

Want more Bloomberg Opinion? OPIN. Or you can subscribe to our daily newsletter.

Like Bloomberg's Points of Return? Subscribe for unlimited access to trusted, data-based journalism in 120 countries around the world and gain expert analysis from exclusive daily newsletters like Markets Daily or Odd Lots.