China’s Economy Has A Few Major Problems

China’s economy is struggling with excessive debt, deflation, excess capacity, and a rapidly aging population. China continues to rely on exports to support economic growth. China has been increasingly accused of dumping its excess production in world markets. This is exacerbating trade frictions, especially with the US.

The Chinese government’s efforts to stimulate domestic consumption have largely failed to achieve this goal. The problem is that Chinese consumers are depressed because many of them are experiencing a significant negative wealth effect from the losses they incurred when China’s property bubble burst. The stock market has also whipsawed them. Consider the following:

(1) New home prices have been falling since mid-2022. They fell 2.2% y/y during September. This marks the 26th consecutive month of decline, reflecting persistent weakness in demand.

(2) The housing slump is dragging down consumer confidence and household spending.

(3) Retail sales in China grew by 3.0% y/y in September 2025, marking the slowest expansion since August 2024. While the overall growth is positive, it reflects a cooling trend compared to the 3.4% increase in August. While some consumer categories are rebounding, others—especially discretionary goods—are losing momentum. Adjusted for the 0.8% y/y drop in China’s CPI for goods, retail sales rose 3.8%. However, this measure has been growing more slowly than industrial production since early last year, which is causing deflation.

(4) The Peoples Bank of China has been providing a stimulative monetary policy by reducing bank reserve requirements and lowering its official interest rate. Yet the y/y growth in bank loans has dropped nearly in half over the past three years to 6.6% y/y.

(5) China’s bank loans rose to a record high of $38.0 trillion in September. That’s a staggering amount of debt, and that is only bank loans.

(6) Both short-term and long-term government interest rates remain depressed below 2.00%.

(7) China’s major stock market indexes have been very volatile and nearly flat for 18 years!

(8) China’s stock market has performed very well this year, with the FTSE China index up 34.7% ytd. Technology (34.8%) has performed well, but even better-performing sectors include Basic Materials (77.7%), Health Care (67.6%), and Consumer Discretionary (48.3%).

Goldman Sachs adds these data:

Household consumption declined sequentially: According to the NBS quarterly household survey, household disposable income grew by 4.5% yoy (3.9% quarter-over-quarter annualized) in Q3, vs. 5.1% yoy (1.9% quarter-over-quarter annualized) in Q2. Household nominal consumption growth measured in year-over-year terms slowed to 3.4% in Q3 from 5.2% in Q2. On a sequential basis after our seasonal adjustment, household consumption per capita in nominal terms decelerated to -2.9% quarter-over-quarter annualized in Q3, vs. an increase of 3.4% quarter-over-quarter annualized in Q2. The deceleration in consumption growth was mainly driven by weaker spending on food, medicine and medical services, and residence.

Which contradicts Yardeni who says discretionary spending has slowed the most.

The labor market softened further: The official urban surveyed unemployment rates rose slightly from 5.1% in Q2 to 5.2% in Q3 after our seasonal adjustments, and the weighted average of employment sub-indexes under various PMI surveys declined in Q3 vs. Q2. Various wage-related indicators showed slower growth in Q3. Specifically, year-over-year growth of official wage income and migrant workers’ monthly average income declined to 4.2% and 2.4% in Q3, respectively, from 4.7% and 3.0% in Q2. The labor cost sub-index in the Cheung Kong Graduate School of Business (CKGSB) Business Condition Index (BCI) survey also showed slower growth in Q3. Our revamped wage tracker suggests urban wage growth moderated further to 3.8% yoy in Q3 from 4.0% yoy in Q2.

Household savings rate ticked up: The household savings rate increased from 31.3% in Q2 to 32.4% in Q3 after seasonal adjustments and rose above pre-Covid trend-implied levels. Our estimated “household excess deposits“, which compares the actual amount of household bank deposits to their pre-Covid trend, reached RMB 57 trillion in Q3. Global historical experience shows that it is hard to change households’ savings behavior, and the most important driver of consumption in China is still likely to be income growth.

Limited impact of consumption boosting policies so far: The NBS consumer confidence index remained depressed in the first two months of Q3 (August as the latest data available). The government has rolled out a few easing measures to boost consumption in recent months, including the nationwide childbirth subsidy, a subsidy program launched in July for elderly citizens with moderate to severe disabilities to support elderly care services consumption in late July, a free pre-school program in August, and a temporary interest subsidy for consumption-related loans in September. However, the macro impact of these programs has been very limited so far.

China’s $1 Billion of Daily US Exports Show Xi’s Bargaining Power

Six months into Donald Trump’s trade war, the resilience of Chinese exports is proving just how essential many of its products remain even after US levies of 55%.

Every day, about a billion dollars worth of goods is crossing the Pacific from China to the US, with the amount ticking up in September from August. Despite double-digit drops in the value of overall trade during the past half a year, some products have recently seen an increase from 2024, defying trade strains between Beijing and Washington.

The upshot is that US tariffs appear somewhat limited in their ability to control what American firms import, as China’s sway over sectors such as rare earths and electronics makes its products hard to dislodge, at least in the short term. That may change over time, especially if Trump further hikes tariffs, as the Republican leader has repeatedly threatened to do.

All that’s giving President Xi Jinping more bargaining power as his trade negotiators head into talks aimed at extending a 90-day tariff truce that’s set to expire in November. In the third quarter, more than $100 billion worth of Chinese goods arrived in the US, helping Beijing keep economic growth on track for its annual target and pushing the bilateral trade surplus up to $67 billion.

Trump on Tuesday predicted an upcoming meeting with his Chinese counterpart would yield a “good deal” on trade, while also cautioning the expected sitdown at a summit in South Korea next week could still fall apart. The US leader has listed rare earths, fentanyl and soybeans as the top trade issues for his side to discuss with China. (…)

While almost all the top 10 exports to the US slumped last quarter from a year earlier, shipments of e-cigarettes rose, according to a Bloomberg analysis of China’s customs data. E-bikes are also seeing strong US demand, with Chinese firms exporting more than $500 million worth in the three months through September, slightly up on a year earlier.

Exports of refined copper cathodes have soared in value terms from almost nothing to $270 million in the past three months, with electrical cables rising 87% to $405 million. (…)

Cracks in Trump’s tariff wall are probably making some of the trade possible by keeping costs down.

ANZ’s Xing said American importers are able to pay a lower levy by declaring the customs value of goods based on their first sale in a third country, and then raising the price when the items reach a US port. Trans-shipping via Mexico or Vietnam means some firms are likely not paying the full tax.

“There are a lot of loopholes,” Xing added. US Customs “just don’t have enough manpower to address them.”

In the July-September quarter, companies in China shipped almost $8 billion worth of smartphones, laptops, tablets and computer parts to the US. While that was less than half the amount sold in the same period last year, it still represented a substantial haul considering the high tariffs.

And despite the end of the “de minimis” rule allowing small parcels to enter the US duty-free, US consumers have kept buying billions of dollars worth of packages from e-commerce platforms such as Shein Group Ltd. and PDD Holdings Inc.’s Temu. While tariffed at 54%, Chinese data showed about $5.4 billion worth of these small packages were sent to the US since the Trump administration closed the loophole in May.

Business to business e-commerce exports also soared, jumping to $201 million in September from about $31 million in August. The surge may indicate Chinese online platforms are moving from selling direct to US consumers to shipping first in bulk and then breaking that down into smaller packages in the US. (…)

There’s been a collapse in exports of games consoles, with companies such as Nintendo Co. and Microsoft Corp. choosing to deliver them from Vietnam and elsewhere instead of paying the higher tariff to ship from China. And US consumers now look to be buying TVs elsewhere, with a 73% drop in the value of LCD sets exported from China to the US last quarter.

Canada: Inflation picks up in September despite a weakening economy

Inflation in September surprised economists by rising two tenths of a percentage point above expectations, with the overall index coming in at 2.4%. Food prices jumped 0.5% in the month, the sharpest increase in six months. On a three months annualized basis, the price of food increased 5.2%, contributing significantly to recent inflation.

Inflationary pressures in September were still fairly widespread, with six of the eight main components rising from August at an annualized rate above the Bank of Canada’s target. Is the September’s CPI report enough to prevent the Bank of Canada from lowering rates at the end of the month? The decision is certainly more complicated following this report, but we believe that it should still favor accommodation.

It is true that total inflation was the highest in seven months (0.4%, m/m), but core inflation measures are much less concerning in September. Deputy Governor Mendez’s recent speech clearly indicated that the Bank of Canada had made another U-turn in regard to their core measures by stating that CPI-Trim and CPI-Median should no longer be prioritized.

For this reason, we will now track these two measures, as well as inflation excluding food and energy and the CPIX (the Bank of Canada’s former core inflation measure). Taking the average of these four measures, the price increase in September was only 0.24%, which is uncomfortable for the Bank of Canada but much less worrying than the increase in the total index. Over three months, the annualized rate of increase is 2.3%, only a few tenths of a percentage point above the target.

This inflation backdrop would be worrying if the economy were showing signs of strength, but this is definitely not the case. Tariff uncertainty continues to weigh on business confidence, with hiring and investment intentions remaining sluggish in light of the business outlook survey. There is therefore a high risk that this economy in excess supply (and too many workers on the sidelines) will become even more so in the coming months.

As for inflationary pressures, there was no indication in the BOS that businesses felt they had pricing power. In fact, on average, anticipated output price growth is similar to what has been seen recently. In conclusion, we continue to favor a rate cut at the next decision, and the need for further accommodation will depend on the federal budget and a potential de-escalation of trade tensions with the United States.

FYI:

There is a positive correlation between the crude oil price and the 10-year US Treasury bond yield (chart). If the oil price continues to fall and the Fed eases on October 29, the yield is likely to fall below 4.00% possibly down to 3.75%. (Ed Yardeni)

A Quality Bull?

Schroeder

AI CORNER

What AI to use in late 2025

Every few months I write an opinionated guide to how to use AI¹, but now I write it in a world where about 10% of humanity uses AI weekly. The vast majority of that use involves free AI tools, which is often fine… except when it isn’t. OpenAI recently released a breakdown of what people actually use ChatGPT for (way less casual chat than you’d think, way more information-seeking than you expected). This means I can finally give you advice based on real usage patterns instead of hunches. I annotated OpenAI’s chart with some suggestions about when to use free versus advanced models.

If the chart suggests that a free model is good enough for what you use AI for, pick your favorite and use it without worrying about anything else in the guide. You basically have nine or so choices, because there are only a handful of companies that make cutting-edge models.

All of them offer some free access. The four most advanced AI systems are Claude from Anthropic, Google’s Gemini, OpenAI’s ChatGPT, and Grok by Elon Musk’s xAI. Then there are the open weights AI families, which are almost (but not quite) as good: Deepseek, Kimi, Z and Qwen from China, and Mistral from France. Together, variations on these AI models take up the first 35 spots in almost any rating system of AI. Any other AI service you use that offers a cutting-edge AI from Microsoft Copilot to Perplexity (both of which offer some free use) is powered by one or more of these nine AIs as its base.

How should you pick among them? Some free systems (like Gemini and Perplexity) do a good job with web search, while others cannot search the web at all. If you want free image creation, the best option is Gemini, with ChatGPT and Grok as runners-up. But, ultimately, these AIs differ in many small ways, including privacy policies, levels of access, capabilities, the approach they take to ethical issues, and “personality.”

And all of these things fluctuate over time. So pick a model you like based on these factors and use it. However, if you are considering potentially upgrading to a paid account, I would suggest starting with the free accounts from Anthropic, Google, or OpenAI. If you just want to use free models, the open weights models and aggregation services like Microsoft Copilot have higher usage limits.

Now on the hard stuff.

If you want to use an advanced AI seriously, you’ll need to pay either $20 or around $200 a month, depending on your needs (though companies are now experimenting with other pricing models in some parts of the world). The $20 tier works for the vast majority of people, while the $200 tier is for people with complex technical and coding needs.

You will want to pick among three systems to spend your $20: Claude from Anthropic, Google’s Gemini, and OpenAI’s ChatGPT. With all of the options, you get access to advanced, agentic, and fast models, a voice mode, the ability to see images and documents, the ability to execute code, good mobile apps, the ability to create images and video (Claude lacks here, however), and the ability to do Deep Research.

They all have different personalities and strengths and weaknesses, but for most people, just selecting the one they like best will suffice. Some people, especially big users of X, might want to consider Grok by Elon Musk’s xAI, which has some of the most powerful AI models and is rapidly adding features, but has not been as transparent about product safety as some of the other companies. Microsoft’s Copilot offers many of the features of ChatGPT and is accessible to users through Windows, but it can be hard to control what models you are using and when. So, for most people, just stick with Gemini, Claude, or ChatGPT.

Just picking one of these three isn’t enough, however, because each AI system has multiple AI models to select. Chat models are generally the ones you get for free and are best for conversation, because they answer quickly and are usually the most personable. Agent models take longer to answer but can autonomously carry out many steps (searching the web, using code, making documents), getting complex work done. Wizard models take a very long time and handle very complex academic tasks. For real work that matters, I suggest using Agent models, they are more capable and consistent and are much less likely to make errors (but remember that all AI models still have a lot of randomness associated with them and may answer in different ways if you ask the same question again.)

Same question asked of a chat model and an agentic one. You can see the chat model answered “off the top of its head” while the agentic model did outside research and checked a lot of assumptions before answering,

For ChatGPT, no matter whether you use the free or pay version, the default model you are given is “ChatGPT 5”. The issue is that GPT-5 is not one model, it is many, from the very weak GPT-5 mini to the very good GPT-5 Thinking to the extremely powerful GPT-5 Pro. When you select GPT-5, what you are really getting is “auto” mode, where the AI decides which model to use, often a less powerful one.

By paying, you get to decide which model to use, and, to further complicate things, you can also select how hard the model “thinks” about the answer. For anything complex, I always manually select GPT-5 Thinking Extended (on the $20 plan) or GPT-5 Thinking Heavy (if you are paying for the $200 model). For a really hard problem that requires a lot of thinking, you can pick GPT-5 Pro, the strongest model, which is only available at the highest cost tier.

For Gemini, you only have two options: Gemini 2.5 Flash and Gemini 2.5 Pro, but, if you pay for the Ultra plan, you get access to Gemini Deep Think (which is in another menu). At this point, Gemini 2.5 is the weakest of the major AI models (though still quite capable and Deep Think is very powerful), but a new Gemini 3 is expected at some point in the coming months.

Finally, Claude makes it relatively easy to pick a model. You probably want to use Sonnet 4.5 for everything, with the only question being whether you select extended thinking (for harder problems). Right now, Claude does not have an equivalent to GPT-5 Pro.

If you are using the paid version of any of these models and want to make sure your data is never used to train a future AI, you can turn off training easily for ChatGPT and Claude without losing any functionality, but at the cost of some functionality for Gemini. All of the AIs also come with a range of other features like projects and memory that you may want to explore as you get used to using them.

The biggest uses for AI were practical guidance and getting information, and there are two ways to dramatically improve the quality of your results for those kinds of problems: by either triggering Deep Research mode and/or connecting the AI to your data (if you feel comfortable doing that).

Deep Research is a mode where the AI conducts extensive web research over 10-15 minutes before answering. Deep Research is a key AI feature for most people, even if they don’t know it yet, and it is useful because it can produce very high-quality reports that often impress information professionals (lawyers, accountants, consultants, market researchers) that I speak to.

Deep Research reports are not error-free but are far more accurate than just asking the AI for something, and the citations tend to actually be correct. Also note that each of the Deep Research tools work a little differently, with different strengths and weaknesses. Even without deep research, GPT-5 Thinking does a lot of research on its own, and Claude has a “medium research” option where you turn on Web Search but not research.

Connections to your own data are very powerful and increasingly available for everything from Gmail to SharePoint. I have found Claude to be especially good in integrating searches across email, calendars, various drives, and more – ask it “give me a detailed briefing for my day” when you have connected it to your accounts and you will likely find it impressive. This is an area where the AI companies are putting in a lot of effort, and where offerings are evolving rapidly.

I have mentioned it before, but an easy way to use AI is just to start with voice mode. The two best implementations of voice mode are in the Gemini app and ChatGPT’s app and website. Claude’s voice mode is weaker than the other two systems. Note the voice models are optimized for chat (including all of the small pauses and intakes of breath designed to make it feel like you are talking to a person), so you don’t get access to the more powerful models this way.

All the models also let you put all sorts of data into them: you can now upload PDFs, images and even video (for ChatGPT and Gemini). For the app versions, and especially ChatGPT and Gemini, one great feature is the ability to share your screen or camera. Point your phone at a broken appliance, a math problem, a recipe you’re following, or a sign in a foreign language. The AI sees what you see and responds in real-time. It makes old assistants like Siri and Alexa feel very primitive.

Claude and ChatGPT can now make PowerPoints and Excel files of high quality (right now, Claude has a lead in these two document formats, but that may change at some point). All three systems can also produce a wide variety of other outputs by writing code. To get Gemini to do this reliably, you need to select the Canvas option when you want these systems to run code or produce separate outputs. Claude has a specialized artifacts section to show some examples of what it can make with code. There are also very powerful specialized coding tools from each of these models, but those are a bit too complex to cover in this guide.

ChatGPT and Gemini will also make images for you if you ask (Claude cannot). Gemini has the strongest AI image generation model right now. Both Gemini and OpenAI also have strong video generation capabilities in Veo 3.1 and Sora 2. Sora 2 is really built as a social media application that allows you to put yourself into any video, while Veo 3.1 is more generally focused. They both produce videos with sound. (…)

Beyond the basics of selecting models, there are a few things that come up quite often that are worth considering:

• Hallucinations: In many ways, hallucinations are far less of a concern than they used to be, as newer AI models are better at not hallucinating. However, no matter how good the AI is, it will still make errors and mistakes and still give you confident answers where it is wrong. They also can hallucinate about their own capabilities and actions. Answers are more likely to be right when they come from advanced models, and if the AI did web searches. And remember, the AI doesn’t know “why” it did something, so asking it to explain its logic will not get you anywhere. However, if you find issues, the thinking trace of AI models can be helpful.

• Sycophancy and Personality: All of the AI chatbots have become more engaging and likeable. On one hand, that makes them more fun to use, on the other it risks making AIs seem like people when they are not, which creates a danger that people may form stronger attachments to AI. A related issue is sycophancy, where the AI agrees with what you say. The reasons for this are complicated but when you need real feedback, explicitly tell the AI to act as a critic. Otherwise, you might be talking to a very sophisticated yes-man.

• Give the AI context to work with. Though memory features are being added, most AI models only know basic user data and the information in the current chat, they do not remember or learn about you beyond that. So, you need to provide the AI with context: documents, images, PowerPoints, or even just an introductory paragraph about yourself can help – use the file option to upload files and images whenever you need, or else use the connectors we discussed earlier.

• Don’t worry too much about prompting “well”: Older AI models required you to generate a prompt using techniques like chain-of-thought. But as AI models get better, the importance of this fades and the models get better at figuring out what you want. In a recent series of experiments, we have discovered that these techniques don’t really help anymore (and no, threatening them or being nice to them does not seem to help on average).

• Experiment and have fun: Play is often a good way to learn what AI can do. Ask a video or image model to make a cartoon, ask an advanced AI to turn your report or writing into a game, do a deep research report on a topic that you are excited about, ask the AI to guess where you are from a picture, show the AI an image of your fridge and ask for recipe ideas, work with the AI to plot out a dream trip. Try things and you will learn the limits of the system.

I started this guide mentioning that 10% of humanity uses AI weekly. By the time I write the next update in a few months, that number will likely be higher, the models will be better, and some of the specific recommendations I made today will be outdated. What won’t change is the fact that people who learn to use these systems well will find ways to benefit from them, and to build intuition for the future.

The chart at the top of this post shows what people use AI for today. But I’d bet that in two years, that chart looks completely different. And that isn’t just because AI changed what it can do, but also because users figured out what it should do. So, pick a system and start with something that actually matters to you, like a report you need to write, a problem you’re trying to solve, or a project you have been putting off. Then try something ridiculous just to see what happens. The goal isn’t to become an AI expert. It’s to build intuition about what these systems can and can’t do, because that intuition is what will matter as these tools keep evolving.

The future of AI isn’t just about better models. It’s about people figuring out what to do with them.

I personally have been using Perplexity Pro and Gemini for a while and my uses just keep increasing. There is a search button on the blog but AI makes it useless. I simply ask Perplexity to search something on Edge and Odds. Easy and quick.

We are currently travelling in Italy and AI is always by us for just about anything. It’s getting better at finding the best ways to get from A to B, find the best prices and how to reserve. And much more. Contrary to what Ethan Mollick says, Perplexity knows where we are and the whole context of our trip. I doubt Perplexity makes money from my $20 monthly subscription.

Voters are split over banning AI data center construction

The rapid proliferation of data centers to fuel the growth of artificial intelligence in the United States has left public officials on both sides of the partisan aisle and energy companies alike blaming the tech industry for the rising cost of consumer electricity.

Public opinion is mixed when it comes to whether such construction should stop, Morning Consult’s new survey shows, as voters — most of whom are inclined to blame AI data centers for rising electricity prices — show signs that they’d be willing to tap the brakes on the industry’s growth as more become aware of the costs. 37% of voters support a ban on the construction of AI data centers in their communities, compared with 39% who oppose it and 24% who are unsure how they feel.

CONSUMER WATCH

Wealthy Americans Are Spending. People With Less Are Struggling.

(…) The divide between rich and poor is hardly new, in Chicago or the rest of the country. But it has become more pronounced in recent months. Wealthier Americans, buoyed by a stock market that keeps setting records, have continued to spend freely. Lower-income households — stung by persistent inflation and navigating a labor market that is losing momentum — are pulling back.

The top 10 percent of U.S. households now account for nearly half of all spending, Moody’s Analytics recently estimated, the highest share since the late 1980s. Consumer sentiment has climbed among high earners but steadily fallen for other groups.

“This isn’t just an inequality story — it’s a macroeconomic story,” said Lindsay Owens, executive director of the Groundwork Collaborative, a progressive policy group. “As the wealthy continue to consume, that’s masking more and more insecurity and instability in the economy under the hood.”

The split is evident across industries. Well-to-do fliers are snapping up pricey seats in first and business class, as airlines struggle to fill the cheaper seats at the back of the plane. Credit card companies are competing to offer ever-more-expensive cards to high earners who are happy to pay the annual fees in return for exclusive perks — while lower-income households are struggling to make minimum payments on their debts.

Even executives at companies that project mass-market appeal are seeing the trend — and in some cases worrying about its implications.

“Visits across the industry by low-income consumers once again declined by double digits versus the prior year period,” Christopher J. Kempczinski, chief executive of McDonald’s, said on a recent earnings call. “This bifurcated consumer base is why we remain cautious about the overall near-term health of the U.S. consumer.” (…)

Slower wage growth, combined with persistent inflation, is straining many families’ finances. Americans are increasingly relying on credit cards and other forms of borrowing to pay their bills, and more are falling behind on car loans and credit card payments.

Those strains have not resulted in widespread defaults, bankruptcies or foreclosures. But high debt balances mean that even people who are keeping up with payments have little room to borrow more if their costs rise or their incomes fall. And data on spending from Numerator, a consumer research firm, shows that lower-income households have cut back on discretionary purchases, leaving them little buffer.

“People are still consuming the basics, but they’re cutting back on all this extra stuff they were able to do coming out of the pandemic,” said Leo Feler, chief economist at Numerator. “It’s just more precarious because if we’ve already trimmed all the fat, the only thing left to trim are the essentials.” (…)

Farmers have been hard hit by Mr. Trump’s trade war with China. Cuts to the federal work force have taken a toll in Northern Virginia and other parts of the country that depend heavily on government employment — effects aggravated by the government shutdown. And immigration raids are weighing on industries that rely on foreign-born workers and on the businesses that count them as customers. (…)

Economists at the Federal Reserve Bank of Boston recently found that growth in consumer spending since 2022 “has been propelled by the highest-income consumers.”

“By contrast,” the researchers noted, “spending growth for low-income consumers has been much weaker.”

The divergence creates two sources of fragility, warned Dhiren Patki, an author of the Boston Fed study. With so much riding on high earners, the economy could suffer if stock prices fall or some other shock leads them to pare their spending. And lower-income households are already stressed financially, leaving them vulnerable if the labor market weakens further. (…)

Nearly two million Americans are considered long-term unemployed, the highest since the pandemic. And joblessness has risen sharply for Black workers, recent graduates and other groups that are often the first to feel the effects of a weakening labor market. (…)

John Authers:

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.

Another K shape chart, this one from Ed Yardeni, showing that the earnings boom is not reaching mid and small cap companies. That’s a lot of companies.

More and more we see signs that there are now 2 Americas. Unhealthy and unsustainable.

First Ford, Now Jeep. Automakers Are Hit by Lack of Parts Supply-chain snafus for rare-earth minerals, aluminum and semiconductors have hit carmakers simultaneously

Assembly lines inside a Michigan factory that churns out high-end Jeep SUVs ground to a halt last week and won’t resume production until early next month. The cause, according to an official for the United Auto Workers, is a shortage of aluminum.

Ford has paused production at three plants for the same reason. Between the two automakers, thousands of workers in Michigan and Kentucky are now collecting unemployment. (…)

All this is rattling an industry that has already been hampered by billions of dollars in tariff payments and a costly pivot away from electric vehicles. (…)

Industrywide, the [tariff] cost exceeds $12 billion. (…)

John Bozzella, chief executive of the Alliance for Automotive Innovation, the top U.S. car industry group, warned that the Nexperia situation could deteriorate quickly and affect the global economy.

“If the shipment of automotive chips doesn’t resume—quickly—it’s going to disrupt auto production in the U.S. and many other countries and have a spillover effect in other industries,” Bozzella said. “It’s that significant.”

Don’t know the “Nexperia situation”? See last Friday’s post Earthquakes.

• U.S. automakers on track to pay $10-billion in tariffs by end of October

The tally comes from a report by Anderson Economic Group (AEG), a consultancy based in East Lansing, Mich., that analyzed data supplied by the U.S. Census Bureau.

Patrick Anderson, principal of AEG, said automakers that have operations that span the North American borders are faced with a “huge bill.” The added costs will drive up car prices, reduce sales and cause autoworker layoffs, he said.

“There is no way for $10-billion to be absorbed by the automakers and suppliers alone,” Mr. Anderson said. “Consumers and workers are going to bear some of these costs.” (…)

Mr. Anderson said the figures underestimate the total tariff costs borne by companies in the U.S. because they include only the two major categories of cars and parts. Also left out are tariffs on steel and aluminum, and imported automobiles from Europe and Asia, he said. (…)

Peter Frise, a professor at the University of Windsor, said Mr. Trump’s goal of forcing automakers to abandon their North American supply lines and ramp up U.S. production faces a tough reality: a U.S. labour shortage.

“They don’t have a lot of highly skilled people, and their education system is not pumping out a lot of highly skilled people,” Prof. Frise said by phone.

“This is something the car companies look at because there’s no point in putting up a plant where you need to employ 10,000 highly skilled people if you can’t hire 10,000 highly skilled people in that locale.”

• Canada needs to start thinking about an auto sector without U.S. automakers

Canada is stuck in an increasingly abusive relationship with the three traditional North American automakers.

And what governments here need to start asking themselves is whether – and if so, how – they can get out of it and maintain a robust domestic auto industry. (…)

In 2007, per data compiled by the Trillium Network for Advanced Manufacturing, the three automakers assembled nearly 1.7 million vehicles in Canada (…). (…) the annual production total is down to around 600,000 cars, in a good year.

(…) while the two Japanese giants have generally been better corporate citizens – maintaining stable and supportive presences in towns where they’ve put down roots, and being less cutthroat in demanding subsidies (even if they still take them when available) – the U.S.-based companies have maintained an outsized squeaky-wheel influence. (…)

So while still engaging with these companies when necessary, this is the time for Ottawa to start considering other long-term strategic options to reduce dependence on them – of which there is no shortage, even if each comes with its share of challenges and obstacles.

The most pressing, though contentious, place to start is with the China question. (…)

Few people in the industry seriously believe Canada will be able to keep out Chinese EVs forever, if they continue to be cheaper and better-made than Western products (not to mention if blocking them hurts other industries because of retaliatory action).

So the question to be asked is whether China would embrace a build-where-you-sell model – possibly involving duty remissions, a mechanism Canada once used to attract Japanese manufacturing, in which tariffs are waived or reduced in return for manufacturing investments.

That may not be something Mr. Carney can immediately embrace, given how much it would poke Mr. Trump, and it’s unclear how much interest China would have in making cars here if it couldn’t access the U.S. market. But it’s a scenario that at least needs to be fully explored.

China’s Economy Grows at Weakest Pace in a Year

• The economic expansion slowed in the July-Sept. period, growing 4.8% (in inflation adjusted terms) from a year earlier. Without those adjustments for changing prices, the economy grew 3.7%, with economy-wide prices falling for a 10th straight quarter
• In September, industrial output rose 6.5%, the best result since June, while retail sales expanded 3% [August +3.4%], the worst for a single month since November last year.
• The housing market worsened, with prices falling more in September than in August. Residential property sales fell 7.6% in the first nine months of the year, putting the nation on track for a third year of declines, and that dragged down property investment, which fell almost 14% for the same period.
• The decline in spending on property is one factor pulling down economy-wide investment, which fell 0.5% so far this year, the first decline for that period since data begins in 1998

Over the first nine months of the year, China’s economy expanded 5.2% from the year-earlier period, according to the National Bureau of Statistics.

US-China Talks Planned Next Week as Trump Plays Down Tariffs

Bessent said he spoke virtually with He on Friday evening. The Treasury chief earlier described the discussions with He as “frank and detailed” and reaffirmed plans to meet in-person next week. US Trade Representative Jamieson Greer also took part in the online talks.

“He and I, and a delegation, will meet in Malaysia — probably a week from tomorrow, to prepare for the two presidents to meet,” Bessent said earlier at a White House event. (…)

“I think we’re doing very well. I think we’re getting along with China,” Trump said. The US president also indicated that he believed his planned meeting with Chinese President Xi Jinping, set to take place this month in South Korea on the sidelines of the Asia-Pacific Economic Cooperation leaders summit, would go ahead.

(…) on Friday, Trump characterized the return of sky-high tariffs as “not sustainable” in a clip of an interview with Fox Business. (…)

“I think things have de-escalated,” Bessent said Friday. “We hope that China will show the respect that we have shown them. And I am confident that President Trump, because of his relationship with President Xi, will be able to get things back on a good course.”

The U.S. Is Tiptoeing Away From Many of Trump’s Signature Tariffs The administration is considering lifting duties on some products not produced in the U.S.

The Trump administration is quietly watering down some of the tariffs that underpin the president’s signature economic policy.

President Trump in recent weeks has exempted dozens of products from his so-called reciprocal tariffs and offered to carve out hundreds more goods from farm products to airplane parts when countries strike trade deals with the U.S.

The offer to exempt more products from tariffs reflects a growing sentiment among administration officials that the U.S. should lower levies on goods that it doesn’t domestically produce, say people familiar with administration planning. That notion “has been emerging over time” within the administration, said Everett Eissenstat, deputy director of the National Economic Council in Trump’s first term. “There is definitely that recognition.”

The move comes ahead of a Supreme Court hearing in early November on the reciprocal tariffs—a case that could force the administration to pay back many of the levies if it loses in court. The White House,

The shift on the reciprocal tariffs reflects the Trump team’s desire to hedge its bets should the court strike down its broad levies, according to people familiar with the administration’s thinking. At the same time, Trump’s team is expanding its use of tariffs based on more established legal authority: Section 232 of the Trade Expansion Act of 1962. Trump has already deployed that law to underpin tariffs on metals and automobiles, and this month announced a new tranche of duties aimed at heavy trucks, pharmaceuticals and furniture.

On Friday, Trump unveiled his latest action under Section 232, imposing 25% tariffs on trucks and truck parts, as well as 10% tariffs on buses, effective Nov. 1. As part of that action, Trump also expanded a tariff relief program for automakers, allowing them to apply for credits to partially offset the cost of tariffs on car and truck parts until 2030, instead of 2027.

Last month, Trump issued new exemptions for products from gold to LED lights and certain minerals, chemicals and metal products via a list called “Annex II” that includes many products that are or will be covered by the Section 232 levies.

He also previewed hundreds of potential exemptions to come in the future: The order includes a list of products that could receive zero tariffs under trade agreements with foreign nations that are being negotiated by Trump’s team. That list, dubbed “Annex III,” is aimed at “products that cannot be grown, mined, or naturally produced in the United States,” the order states, such as “certain agricultural products; aircraft and aircraft parts; and non-patented articles for use in pharmaceutical applications.”

The September order also allows new authority to the Department of Commerce and the U.S. Trade Representative’s office to grant tariff exemptions themselves, without Trump himself issuing executive orders mandating the new carve-outs.

The move will help streamline tariff policy, a White House official said, so the administration doesn’t need to issue an executive order for every group of exemptions as it implements over a dozen trade deals Trump has announced, or arrives at new pacts. (…)

For months, administration officials led by Commerce Secretary Howard Lutnick had insisted there would be “no exemptions, no exceptions” from Trump’s so-called reciprocal duties, originally announced in April. Lutnick has softened his stance publicly, saying in a late July television appearance that “if you grow something and we don’t grow it, that can come in for zero.” (…)

EARNINGS WATCH

58 companies in the S&P 500 Index have reported earnings for Q3 2025. Of these companies, 86.2% reported earnings above analyst expectations and 12.1% reported earnings below analyst expectations. In a typical quarter (since 1994), 67% of companies beat estimates and 20% miss estimates. Over the past four quarters, 77% of companies beat the estimates and 18% missed estimates.

In aggregate, companies are reporting earnings that are 6.4% above estimates, which compares to a long-term (since 1994) average surprise factor of 4.3% and the average surprise factor over the prior four quarters of 7.1%.

Of these companies, 81.0% reported revenue above analyst expectations and 19.0% reported revenue below analyst expectations. In a typical quarter (since 2002), 62% of companies beat estimates and 38% miss estimates. Over the past four quarters, 67% of companies beat the estimates and 33% missed estimates.

In aggregate, companies are reporting revenues that are 1.5% above estimates, which compares to a long-term (since 2002) average surprise factor of 1.3% and the average surprise factor over the prior four quarters of 1.5%.

The estimated earnings growth rate for the S&P 500 for 25Q3 is 9.3%. If the energy sector is excluded, the growth rate improves to 10.1%.

The estimated revenue growth rate for the S&P 500 for 25Q3 is 6.2%. If the energy sector is excluded, the growth rate improves to 6.8%.

The estimated earnings growth rate for the S&P 500 for 25Q4 is 7.8%. If the energy sector is excluded, the growth rate improves to 8.1%.

Most of the reporters last week were Banks. The largest banks generally highlighted strong trading and M&A activity and a healthy state of the consumer. However, some regional bank reports raised concerns over credit quality. 14% of S&P 500 market cap will report next week.

Trump’s economic adviser: AI productivity surge is real, won’t cause inflation

The AI-driven productivity surge is real, it’s already underway, and this will allow robust growth without inflation, a top White House economic adviser and potential next Federal Reserve chair argues.

It’s a signal that at high levels of the Trump administration, the investment surge driven by AI is viewed as an unalloyed positive, as it will mean an ongoing jump in the economy’s productive potential — not unlike in the late 1990s.

“If you have … an understanding of how economies work and you understand that because of AI we’re in the midst of a huge positive supply shock,” that pushes supply up in the economy and prices down, Kevin Hassett, director of the White House National Economic Council, said at an Axios News Shapers event Wednesday morning.

• “And so it’s actually quite a deflationary thing, and there was precedent for this. I worked at the Federal Reserve in the ’90s with Alan Greenspan, and you might recall that there was a time when the computer revolution was starting to happen, equity markets were celebrating. The unemployment rate was really, really low.”
• “Greenspan decided that if you’re getting a positive supply shock, then you can let low unemployment stay without having to lift rates, and he gave us three or four really great boom years. … I think that the Fed has an opportunity like that right now.”

In light of those productivity gains, he seemed unconcerned about the risks that there could be a bubble in AI, data centers, and related investments.

• “My belief is that you don’t have to think about something popping, or don’t have to think it’s a high probability because the productivity gains are so high,” Hassett said.

Hassett is said to be one of five finalists for the Fed chief job when chair Jerome Powell’s term is up in May.

Mr. Market has been avoiding the “productivity play” so far, at least based on Goldman Sachs’ Productivity basket. The current focus is all on the Build and Energy parts.

The WaPo got an opinion piece from Fabien Curto Millet, Google’s chief economist and Diane Coyle, Bennett professor of public policy at the University of Cambridge. It begins with (my emphasis):

Is artificial intelligence going to destroy jobs, as some recent research warns?

Some labor market disruption is certainly a real possibility. But the true challenge for governments and businesses is ensuring that workers have the skills and adaptability needed to use the technology. Only then will AI drive productivity and raise living standards. (…)

This is where the key challenge lies now — and here the lessons of history are less encouraging. In many prior waves of technology, such as the automation that took place across manufacturing in the 1980s and ’90s, support for workers needing to change how they worked or to find other occupations was inadequate. Many who were laid off were unable to find new jobs, and their communities still bear the scars. Policymakers should be thinking now about how to manage the transition better this time.

The key is training workers to make the most of AI. Since most of our 2030 workforce is already employed, we should meet people where they are and provide opportunities to acquire new skills mid-career. There is economic evidence that appropriately designed reskilling support can be highly effective. (…)

One real AI impact:

Is the AI revolution raising consumers’ utility bills?

The average utility payment for electricity and gas rose 3.6% year-over-year (YoY) in 3Q 2025, according to Bank of America internal deposit data. This was down from a peak rise of nearly 10% YoY in the spring, providing some respite to consumers perhaps.
However, it’s likely that this reprieve is due partly to relatively soft electricity demand over the summer rather than energy prices.

In fact, electricity and gas prices, as measured by the Bureau of Labor Statistics’ (BLS) consumer price index, show YoY price increases of 6% and 14%, respectively, in August. So, in our view, consumers may again feel the pressure on their utility bills in the coming months, particularly if the winter is a cold one. (…)

BofA Global Research argues that rising demand for electricity from both data center development and manufacturing growth is already being reflected in residential customer rates.

The impact runs through the spending on enhancements to the transmission and distribution grid required for data center build-outs, which is incorporated into the tariffs of all the ratepayers (residential, commercial and industrial) on the system, and then into both higher energy and capacity pricing.

Capacity prices are charged to ensure future electricity demand will be reliably met and BofA Research points to capacity auctions in the Pennsylvania-New Jersey-Maryland Interconnection (PJM) region as an example of rising cost pressure. PJM is a regional transmission organization (RTO) covering around 65 million customers in the eastern US. It hosts an open and competitive auction process to set the price for capacity in the 13 states where it is responsible for electricity dispatch and system reliability.

PJM’s capacity auctions have seen five-fold increases in prices: the auctions for delivery in June through May rose from $34/MegaWatt (MW)-day in 2023/24, to $269/MW-day in 2025/26. For the 2026/2027 delivery year auction, capacity prices cleared at $329/MW-day reflecting a cap that was agreed to by the Governor of Pennsylvania and PJM. Pricing would have been $388/MW-day without that cap.

These rising capacity rates impact residential customers because, although the rates that utilities charge customers are regulated, utilities are entitled to fully recover some expenses – capacity prices and the price of power are chief among them.

How will the growth in electricity demand impact consumer bills from here? In BofA Global Research’s view, there is likely further upside ahead. They point to the fact that electricity supply is still struggling to catch up with the rapid increases in demand because of the capital intensity and regulatory requirements around building more generation and transmission capacity.

It is peak demand where the supply-demand picture is really tightening. One issue is that the speed at which new supply can be built is also being limited by supply chain pressures, for example in very large turbines. And, while some of peak demand can be met in the short term with solar and storage, there is also regulatory uncertainty around the supply chain in these areas too. So, for now, it appears that these pressures are not over.

For consumers, this is, of course, not welcome news. Higher utility payments tend to disproportionately impact lower-income households more than others. While lower-income households with incomes below $50K have average utility bills around 80% of the US average, households with incomes above $150K have bills around 134% of the US average. So, while higher-income households pay more for utilities, the rise is not proportional to their income. Moreover, relative to their total consumer spending, Bureau of Labor Statistics (BLS) data suggests electricity and gas represented around 4.5% of lower income households’ spending in 2023, compared to 3% across all households.

At a time when lower-income households are already under pressure from slowing wage growth (see: Consumer Checkpoint: The tale of two wallets), rising electricity and gas bills would be another headwind. But, more broadly, rising utility bills could be a headwind to overall consumer discretionary spending if rises are significant and persistent.

Costs of electricity and natural gas have been rising steadily. They are up 15-20% since 2020 in spite of flat natural gas spot prices. This could become election issues in some states.

Coming to a WalMart near you:

Source: J.P. Morgan

It’s not only imported goods:

How mega batteries are unlocking an energy revolution Vast battery units are shoring up grids and extending the use of clean power

(…) Global [battery storage] capacity is expected to rise by 67 per cent to 617GWh this year and to grow tenfold by 2035, according to energy research firm BNEF.

The US and China dominate the market, accounting for about 70 per cent of projects by power capacity. Batteries are also playing a growing role in Australia and the UK and countries from Chile to Saudi Arabia are announcing plans to ramp up deployment.

China has more battery projects in the pipeline than the rest of the world combined.

(…) the global average price of battery storage systems have more than halved over the past two years.

Lithium-ion battery costs have also fallen — by 90 per cent since 2010 — a drop that Artem Abramov, deputy head of research at energy consultancy Rystad, says is likely to continue.

Packed full of hundreds of powerful batteries, a standard 20-foot storage unit once provided 3-4 megawatt hours (MWh). Now they typically deliver 5-6MWh, with several suppliers developing 10MWh containers — enough to power around 30,000 UK homes for an hour.

This innovation has ushered in a wave of battery farms. In 2022, there was only a single gigawatt-scale facility — defined as having a capacity of at least 1GWh, able to supply roughly 3mn UK households for an hour — in operation worldwide. Today there are 42 such sites.

Five times as many giga-projects are set to come online in the next couple of years, including in the UK, the Netherlands, Chile and the Philippines.

Last month, Tesla — the only non-Chinese company in the top five battery storage system manufacturers globally — unveiled Megablock, a large-scale battery system that the company says will significantly cut installation times and reduce construction costs.

Megablock will be built and pre-packaged within a Tesla factory, with a design that reduces the number of electrical connections. Mike Snyder, vice-president of energy and charging, said at a launch event in Las Vegas that Megablock would allow Tesla to install 1GWh in 20 business days, the equivalent of “bringing power to 400,000 homes in less than a month.”

A week later, Chinese EV giant and battery maker BYD announced its new Haohan battery system, which has twice as much power as a Tesla battery block when adjusted for size.

China, which produces over three-quarters of batteries sold globally, has played a crucial role in the energy storage boom. (…)

In the US, Trump’s aggressive pushback against renewable energy is threatening to slow the sector’s momentum. But analysts anticipate a combination of the significant drop in battery costs and the need to store energy means that projects will still get built.

Battery storage capacity in the US jumped 63 per cent in the first half of this year, according to data from S&P Global, driven by a surge in Texas. (…)

“The combination of solar and batteries means clean power costs in sunbelt regions could be 50 per cent lower than today’s fossil-based systems.”

In countries more reliant on wind power — such as the UK, Germany and Canada — batteries alone will not suffice, according to the ETC. Wind is less consistent, and current battery technology cannot fill the gap over multiple days. (…)

In California, batteries consistently supplied more than a quarter of electricity during this year’s spring and summer peaks, data from energy analytics platform Grid Status shows. In the first eight months of 2025, gas generation was down 37 per cent from 2023, according to climate academic Mark Jacobson of Stanford University.

In Australia, several battery farms are being installed at the sites of retired coal plants. The country’s grid-scale battery storage power capacity has more than tripled since the start of 2024 and in early May, batteries contributed more than 5 per cent of power in the evening peak for the first time. Many countries are hoping to see similar results.

Electricity is “the ultimate perishable good”, Ember’s Walter says, and batteries can unlock its potential much like “grain silos stabilised harvests and refrigeration preserved fresh food. The technology to do this may be new, but the economics are as old as civilisation.”

FYI: Wholesale turkey prices are now about 40% higher than last year, according to the American Farm Bureau Federation. (Axios)

From Callum Thomas:

• Technical Check: After the latest trade war scare selloff, the S&P500 has found support and for now managed a small bounce (and 50-day breadth ticked up from slight oversold levels). While it’s anyone’s guess as to what the next headline or tweet will bring, zooming out to the big picture — with the 50-day moving average holding, the market finding short-term support, and the 200-day upward sloping …it’s still by and large a healthy bull market picture.

Source:  MarketCharts

• Seasonal Scintillations: Again, while we could easily get torpedoed by any number of (geo)political headlines and issues, from a seasonality standpoint we’re heading into what has historically been the better part of the year for markets.

Source:  Topdown Charts

• Cash is Trash — Fund Managers: And as such, fund managers are running very low allocations to cash. Definitely a sign of the times… and they’re not the only ones.

Source:  @KobeissiLetter

• Cash is Trash — Private Wealth: While it may not necessarily be representative, BofA private wealth clients are running multi-year low allocations to cash. That’s multiple datapoints saying the same thing (check out the next chart too).

Source:  @MikeZaccardi

• Cash is Trash — Retail & Households: The AAII survey of retail/individual investors likewise shows cash allocations at the low end of the historical range. The aggregate across US households (Fed Flow of Funds data) also shows allocations lower than average. The only outlier is the ICI implied allocations data, but that is skewed up because during the 2023 regional bank crisis and SVB collapse there was a big migration out of bank deposits and into money market funds (in effort to avoid counterparty risk).

Source:  Topdown Charts Professional

• The All-in Age:  The flipside of low cash allocations (and low bond allocations too, for that matter) is household equity allocations floating around record highs (with valuations likewise soaring to lofty heights). Only time will tell how much further we can take this before things come undone.

Source:  Topdown Charts Research Services

Dry powder no more?

• Banks tap Fed lending facility in sign of short-term market strains Institutions borrow $15bn from standing repo facility in space of two days in sign of liquidity pressures

Banks tapped the Federal Reserve’s short-term lending facility for more than $15bn over the past two days, in a sign of the liquidity pressures in the repo market that could drive the Fed to stop shrinking its balance sheet.

Banks borrowed $6.75bn on Wednesday and $8.35bn on Thursday from the Fed’s standing repo facility (SRF), the largest amount borrowed over a two-day period since the Covid-19 pandemic.

The SRF was introduced in 2021 as a permanent replacement for emergency repo operations launched by the Fed in the wake of market turmoil two years earlier. It is intended as a safety valve that helps limit how high borrowing rates in the short-term funding market can go, by allowing banks to borrow cash from the Fed during twice-daily operations in return for collateral such as US Treasury debt.

In normal conditions, the facility is rarely used, because banks can typically borrow at better rates in the repo market than at the SRF.

But traders and analysts said that banks were forced to borrow at the Fed because repo lending rates rose above the 4.25 per cent offered.

Settlements of Treasury bills and bonds this week contributed to the strain in lending markets. However, those are regular events that have not had the same kind of effect on the market in recent years.

The jump in repo rates suggests that the Fed’s effort to shrink its balance sheet by shedding Treasury debt and mortgage-backed securities — otherwise known as quantitative tightening or QT — is nearing its end, depleting banks’ excess reserves. (…)

“I think that there are now clear signs that the Fed has likely over-drained liquidity from the system,” said Mark Cabana, head of US rates strategy at Bank of America. “QT will end by the end of the year, but it could end even earlier.”

“This is a loss of control of money markets,” said Cabana.

Officials have signalled quantitative tightening could end as soon as this year. Jay Powell, Fed chair, said on Tuesday that he expected to be halting QT in the “coming months”. Powell also confirmed that officials would like banks to continue to hold what they describe as “ample reserves” — code for lenders having enough cash that they are not usually reliant on facilities such as the SRF.

The Fed’s weekly balance sheet showed lenders held $3tn in reserves at the US central bank. “Some signs have begun to emerge that liquidity conditions are gradually tightening, including a general firming of repo rates along with more noticeable but temporary pressures on selected dates,” Powell said.

Many on the Federal Open Market Committee, the central bank body that makes decisions on quantitative easing and tightening, want to avoid a repeat of a scenario where liquidity dries up to an extent that they no longer feel they have control over short-term interest rates.

BUBBLE WATCH

Goldman Sachs:

The large AI hyperscalers have increased their capex investments from $153 billion in 2023 to an estimated $390 billion in 2025. Consensus estimates imply YoY capex growth will decelerate to 20% in 2026, but we believe these estimates are too conservative.

Capex growth registered 78% YoY in 2Q, firms continue to message that supply cannot keep pace with AI demand, and with some exceptions, the AI hyperscalers have generally been able to fund these capex plans via cash flow generation and existing cash balances rather than debt financing.

If the hyperscalers continue to believe the returns on AI investment are compelling enough, they have substantial capacity for more investment spending.

The AI hyperscalers now account for 27% of S&P 500 capex, and we expect their capex spending will exceed the consensus forecast of 20% growth next year.

Whatever ROI analysis the hyperscalers have done are all based on several assumptions that only time will prove or disprove. They are all involved in a race to insure an “adequate” presence in the AI space, fed by investors also tripping on themselves to participate and by provisions in the One Big Beautiful Bill Act that makes R&D, equipment capex, and some structure capex fully deductible expenses.

Source: Carson Group

Goldman’s outstanding research department just released a comprehensive analysis of AI demand with the following estimated demand for chips:

• AI training servers – Full rack AI servers started shipment in 4Q24 and we expect it will ramp up to a larger volume from 3Q25. We now model shipments of 19k/ 50k/ 67k racks in 2025-27E measured in 144-GPU equivalent for a Total Addressable Market size of US$54bn, 157bn and 232bn in 2025-26-27 respectiively.
• AI training servers – High power in stronger ramp up. Shipments will amount to 637k/ 732k/ 819k units in 2025-27E measured in 8-GPU equivalent, or +7%/ +15%/ +12% YoY, with a TAM size of US$136bn/ 138bn/ 139bn.
• Inferencing servers – We expect shipments will increase to 441k/ 522k/ 646k in 2025-27E, or +10%/ +18%/ +24% YoY, with a TAM size of US$27bn/ 30bn/ 36bn in 2025-27E. The growth is driven by increasing applications.
• General servers in gradual recovery. We expect volumes to grow 9%/ 4%/ 3% in 2025-27E and revenues to grow 35%/ 5%/ 4% YoY, supported by the recovery of replacement cycles and the introduction of new CPU platforms.

For now, growth is highest in Full Rack Servers (NVDA).

In all, GS sees the total TAM at $217B in 2025, rising 50% to $325B in 2026 and 31% to $427B in 2027.

Inferencing servers are tiny in GS chart, yet they represent the future of AI. I can’t really comment on the training server assumptions but I venture to say that the inference forecasts will prove too low when applications explode.

The big unknowns are pricing, and …China.

Last week, in the FT:

OpenAI books about $13bn in annual recurring revenue, 70 per cent of which comes from consumers using ChatGPT, which costs $20 for a standard subscription, according to people familiar with the company’s finances. ChatGPT has more than 800mn regular users, but just 5 per cent of those are paying subscribers, a number OpenAI intends to double, the senior executive said.

The company has also rolled out cheaper access to users in India, with plans to do the same in the Philippines, Brazil and elsewhere, they said. (…)

OpenAI’s operating loss in the first half of the year was about $8bn, even as revenue more than doubled on the year before, said a person with knowledge of the matter.

Basic math:

ChatGPT revenues are running at $9.1B annualized from 40 million paying customers = $227.50/yr or $19/m.

760 million people are free riders using compute capacity.

OpenAi has pledged to spend $1T in the next 5 years = $200B/yr

Doubling the % paying to 10% = roughly $20B/yr in revenues. Say its 10% on 1.5T users (!!!) = roughly $40B/yr.

That’s a wide gap to bridge with potential “government contracts, shopping tools, video services, consumer hardware, and even becoming a computing supplier itself through its Stargate data center project.”

Unless many more users opt to pay the $200/m pro rate. That could bring in up to $75B/yr if 20% do. = $115B/yr. Still very short, not even factoring in costs… (If OpenAi lost $8B in the first half, or $16B annualized, this means its annualized costs are running at $25-30B/yr on $13B in revenues.)

… nor competition:

Anthropic,” the Bezos-backed ChatGPT rival, is on track to hit an annual revenue run rate of $9 billion by the end of the year. It was also reported that the Claude maker is almost tripling its annual revenue goals for 2026, which could rise to nearly as much as $26 billion.”

And there’s Google Gemini, Meta AI and Perplexity.ai, etc…

Not counting the cheaper Chinese models.

There’s a whole ecosystem built on OpenAi’s success.

But my basis math is probably too basic, too simple. Otherwise, this is really unhealthy and unsustainable.

Blackstone says Wall Street is complacent about AI disruption Jonathan Gray says private capital group has put risks from the technology ‘top of our list’ when evaluating deals

Wall Street investors are underestimating artificial intelligence’s potential to make entire industries obsolete, Blackstone’s president has said, adding that the impact of the technology was now “top of our list” when evaluating deals.

Jonathan Gray said that understanding AI risks has become a priority for the private capital group when assessing investments, with the technology already upending business models and causing job losses. (…)

Gray said investor exuberance meant it was inevitable that there would be some misallocation of capital to AI companies — “think of Pets.com in 2000”. But he added that the scale of the technology’s impact meant investors may still underestimate its potential to crush entire industries.

“If you think about rules-based businesses — legal, accounting, transaction and claims processing — this is going to be profound,” he added. (…)

Gray said Blackstone had elevated AI risks to the “top of our list” when assessing the potential downside of investments.

“(…) what does AI mean for enterprise software, for service businesses handling data and for rules-based work?” he added.

The rise of AI algorithms created by OpenAI, Microsoft and Google is already disrupting white-collar sectors such as accounting, consulting and law, and threatening business models of companies such as advertisers, publishers and software groups.

Machine -learning technology is also threatening manual jobs in areas such as manufacturing. 

Blackstone (…) has recently decided not to buy some software and call-centre companies seen as vulnerable to AI-related risks, according to people briefed on the matter.

Blackstone has also invested heavily in utility companies that power data centres, even repositioning some of its industrial portfolio companies such as Copeland and Legence to sell products to providers of AI infrastructure. (…)

Gray said that while AI would create some negative economic disruptions, the technology could also yield underestimated productivity benefits for large corporations and the global economy, creating trillions of dollars in new corporate wealth.

So he has challenged dealmakers to also not miss AI-related opportunities. “We’re forcing the conversation. We don’t claim to know exactly how it all plays out. But if every deal team has to analyse AI impact then it’s the number-one topic in the room,” he said. 

“Acting like it’s business as usual would be a mistake,” he added.

Pentagon press turns in badges

Nearly every Pentagon reporter, from almost every major media outlet, turned in their press badges yesterday after refusing to sign onto the Defense Department’s new rules for journalists, Axios’ Josephine Walker writes.

Media companies broadly rejected the pledge, claiming it would criminalize national security reporting and expose those who sign the contract to potential prosecution.

The Pentagon Press Association said its members “are still committed to reporting on the U.S. military. But make no mistake, today, Oct. 15, 2025, is a dark day for press freedom.”

Chief Pentagon spokesman Sean Parnell said in a statement to Axios: “This has caused reporters to have a full-blown meltdown, crying victim online. We stand by our policy because it’s what’s best for our troops and the national security of this country.”

What’s in the new rules (N.Y. Times gift link)

Bloomberg:

President Donald Trump on Friday announced the latest targeted strikes on civilians by the US military, which has repeatedly blown up boats in the Caribbean in what legal experts have called extrajudicial killings that violate military, US and international law. Trump and the Department of Defense have not provided any evidence that the more than two dozen civilians killed by American forces posed a threat to them or were, as the White House has claimed, transporting drugs. The latest strike follows the announcement Thursday that Admiral Alvin Holsey, the head of US Southern Command, plans to retire at the end of the year. Holsey was just one year into his term and no reason was given.

Also on Friday, Trump asked the US Supreme Court to reverse two lower courts and let him deploy soldiers to Chicago streets as part of his ongoing effort to partially militarize Democratic-run cities, including Washington and Portland, Oregon. “Trump will keep trying to invade Illinois with troops—and we will keep defending the sovereignty of our state,” Illinois Governor JB Pritzker said in response. “Militarizing our communities against their will is not only un-American but also leads us down a dangerous path for our democracy.”

And you probably noticed that he made sure the military was paid during the shut down.

I don’t know why, but having just read this from Winston Churchill, who knows fascism better than anybody, I felt like posting it FYI:

Tyranny is our foe. Whatever trappings or disguise it wears, whatever language it speaks, be it external or internal, we must for ever be on our guard, ever mobilized, ever vigilant, always ready to spring at its throat.  (September 6, 1943)

Trump urged Zelenskyy to accept Putin’s terms or be ‘destroyed’ by Russia US president tossed aside maps of Ukraine frontline in volatile White House meeting

(…) The meeting between the US and Ukrainian presidents descended many times into a “shouting match”, with Trump “cursing all the time”, people familiar with the matter said.

They added that the US president tossed aside maps of the frontline in Ukraine, insisted Zelenskyy surrender the entire Donbas region to Putin, and repeatedly echoed talking points the Russian leader had made in their call a day earlier. (…)

During Friday’s meeting, Trump appeared to have adopted many of Putin’s talking points verbatim, even when they contradicted his own recent statements about Russia’s weaknesses, said European officials briefed on the meeting.

According to a European official with knowledge of the meeting, Trump said to Zelenskyy that Putin had told him the conflict was a “special operation, not even a war”, adding that the Ukrainian leader needed to cut a deal or face destruction.

The official said that Trump told Zelenskyy he was losing the war, warning: “If [Putin] wants it, he will destroy you.”

At one point in the meeting, the US president threw Ukraine’s maps of the battlefield to one side, the official familiar with the encounter said. According to the official, Trump said he was “sick” of seeing the map of the frontline of Ukraine again and again. “This red line, I don’t even know where this is. I’ve never been there,” Trump said, according to the official.

Trump also said that Russia’s economy is “doing great”, the official said, in a sharp contrast to his recent public remarks in which he urged Putin to negotiate because his “economy is going to collapse”. (…)

Trump told Fox News on Sunday that he was confident about securing an end to the conflict, and added that Putin is “going to take something, he’s won certain property”.

Putin made a new offer to Trump on Thursday under which Ukraine would surrender the parts of the eastern Donbas region under its control in exchange for some small areas of the two southern frontline regions of Kherson and Zaporizhzhia. The Russian proposal marks a small concession from that made during Putin’s last meeting with Trump in Alaska in August, where he said he would agree to freeze the line of contact elsewhere on the frontline if Ukraine surrendered the Donbas. (…)

But ceding the remainder of the Donbas still under Ukrainian control would be a non-starter for Ukraine, as it would hand Moscow territory it has only partially occupied for more than a decade and failed to seize since Putin ordered the invasion in 2022. Russian forces have struggled to retain the territory in Kherson and Zaporizhzhia that Putin offered in exchange, and have made virtually no progress on the battlefield there since 2022, the year the war began. (…)

Trump’s belligerent repetition of Putin’s rhetoric on Friday dashed hopes among many of Ukraine’s European allies that he could be convinced to increase support to Kyiv. That hope had risen after Trump in recent weeks expressed frustration and impatience with the Russian president’s refusal to engage in bilateral peace negotiations with Zelenskyy.

Three other European officials briefed on the White House discussions confirmed that Trump had spent much of the meeting lecturing Zelenskyy, repeating Putin’s arguments about the conflict and urging him to accept the Russian proposal.

“Zelenskyy was very negative” following the meeting, according to one of the officials, adding that European leaders were “not optimistic but pragmatic with planning next steps”. In a statement on Sunday, Zelenskyy said “decisive steps are needed from the United States, Europe, the G20 and G7 countries” to end the war.

Seems like a playbook, no?