In partnership with


500
Adult male Alaskan brown bears can gain 500 pounds or more before hibernation. (It’s Fat Bear Week.)

AI spend is booming, but demand may not keep up
Markets look strong, but households look weak
Why Palantir just dropped a new line of merch
The Hidden Fragility of AI’s Biggest Deals
The AI infrastructure build-out accelerated last week with hundreds of billions in new commitments.
Nvidia announced plans to invest up to $100 billion in OpenAI. The following day, OpenAI, Oracle, and SoftBank unveiled five new data center sites for their $500 billion Stargate project. One day later, Oracle said it would raise $15 billion in debt to help fund its massive data center build-out — just weeks after OpenAI signed a $300 billion compute deal with the company slated to start in 2027. Finally, CoreWeave expanded its cloud supply agreement with OpenAI by another $6.5 billion, marking the third major increase in the past year.
These deals raise two distinct concerns. In some cases, firms are making commitments with money and assets that don’t yet exist.
OpenAI is promising Oracle $300 billion — money it doesn't have — for infrastructure Oracle hasn’t built.
In others, revenue comes from “circular financing,” where dollars just rotate between companies, obscuring true market demand.
For example, most of the $100 billion Nvidia invested in OpenAI last week, OpenAI will use to buy Nvidia chips.

The concern is that vendor financing and debt could mask weak demand, driving excessive AI build-outs. Until now, most data center spending came from Big Tech’s cash flow, keeping risk contained.
That’s shifting: OpenAI has committed money it doesn’t have to Oracle, and Oracle’s cash flow turned negative this year for the first time since 1992. Reliance on external financing pushes risk beyond shareholders to banks, bondholders, and credit markets. If demand doesn’t keep pace, debt could stretch the AI bubble to its breaking point.


This isn’t the first time we’ve seen massive infrastructure build-outs fueled by overexcitement about world-changing technology.
During the dot-com boom, billions were spent building fiber optic cables for internet demand that hadn’t arrived yet. In the 1870s, railroads absorbed 20% of U.S. capital investment (AI today is 2%), only to collapse into bankruptcy when too many lines failed to generate traffic. In the 1920s, utilities massively overbuilt the electric grid, setting the stage for financial wreckage after the 1929 crash. Each time the technology was real, but the supply and demand timelines were mismatched.
AI is showing familiar symptoms. OpenAI is promising Oracle $60 billion a year for compute it doesn’t use, on infrastructure Oracle hasn’t built, powered by energy it doesn’t have, running on chips from Nvidia — which is investing $100 billion into OpenAI. Microsoft, Oracle, Nvidia, OpenAI are all simultaneously one another’s customers, investors, and suppliers.
The critical difference today is exposure. In the railroad era, less than .1% of Americans owned stock or railroad bonds. Compare that with today: 62% of Americans own stocks whether directly or through mutual funds or retirement accounts. A much larger percent of households is exposed to the market.
The question isn’t whether AI is transformative — it almost certainly is. The question is whether the financing loops can hold long enough for revenues to catch up.

I believe there can only be one winner here — an agentic layer controlled by a single company that seamlessly powers everything. If you buy into that, this spending looks rational.
But the sheer scale of these commitments has convinced me of one thing: OpenAI is going to try and go public in the next 12 months. It won’t be able to raise $600 to $800 billion privately, but they’ve built an unbelievable brand. I wouldn’t be surprised if they announce what will be the biggest IPO in history.
Headline Numbers Hide Economic Weakness
By the biggest economic indicators, the economy looks resilient. Second quarter GDP grew 3.8%, its fastest growth pace in nearly two years. Corporate earnings are on track to rise 12% from last year — the third straight quarter of double-digit growth. The unemployment rate reached 4.3% in August, well below historical recession thresholds. Retail sales rose last month. Indexes hit all-time highs last week.
However, those headline numbers are a little misleading. Earlier this month we talked about how the U.S. economy has become winner-take-all. The top wealthiest 10% of Americans now account for a record 50% of all consumer spending; the top 10 U.S. stocks are worth more than China’s entire stock market and have accounted for 55% of the index’s gain since 2021.
The degree to which the top echelons of people and stocks control the economy and markets isn’t just problematic from a fairness perspective. It distorts our understanding of the broader economy: Concentrated gains among a few households and stocks can make the economy look stronger than it is. This misalignment can lead policymakers, investors, and business leaders to make decisions based on aggregate numbers that don’t reflect underlying reality.
The reality is that, by many metrics, middle- and low-income households are struggling.
Sales of very cheap food items — foods people buy when times are tough like rice, canned tuna, beans, macaroni — have risen significantly since the beginning of the year. Hamburger Helper sales are up 15% through August.
On Yelp, searches for “cheap eats” are up 21% compared with last year.
This divide is showing up in the housing market, too. Google searches for "help with mortgage" have surpassed levels last seen during the Great Recession. FHA mortgages, which are typically used by first-time buyers or lower-income individuals, now account for nearly 40% of delinquencies, despite making up just 12% of mortgage balances. That is up from 30% before the pandemic.

Also: health care. One-third of adults in the past year say that they’ve opted out of getting the health care they needed because of the cost. That’s up 9 percentage points since 2023.
Lesser-known economic indicators are also flashing warning signs. By the end of September, U.S. cardboard box-making capacity will be down about 9% in just eight months — roughly double the cuts seen in the 2009 recession. Concerningly, this decline is coming right before the holiday season, when people are usually buying and shipping more stuff.
In past downturns, economic stress led to more pets ending up in shelters. The same is happening today. In Charlotte, North Carolina, surrenders are up nearly 43% this year. Chicago shelters are now taking in an average of 56 animals a day, up from 42 last year. And in New York City, three Animal Care Centers have stopped accepting new animals altogether.
One more data point: Investopedia recently estimated the cost of the American Dream — a wedding, home, kids, college, etc — and it came out to more than $5 million. That’s more than the $2.8 million the average American with a bachelor's degree will make over the course of their career. Achieving the American Dream requires, at the very least, a dual income, college-educated home.
That might not sound like a big deal, but households with two incomes where both earners have at least a bachelor’s degree represents only about 7% of American households.
The American Dream has become the American pipe dream.

The sad part is, America knows what’s going on, and we’re down with it. Our superpower is that we believe someday we’re gonna be in that top 10, maybe that top 1%, so we’re not as worried about the bottom 90.
How does this end? You could see a pretty big economic unwinding. You’ll also see more Mamdani-like candidates. What do I mean by that? We don’t have a tendency to be able to stop the pendulum in the middle. We have a tendency to swing very hard the other way.
I think that it’s likely, assuming we have another free election, you’re going to see an overcorrection to the left. The algorithms don’t like moderates. I think the correct reaction would be ending up with someone in the middle, but unfortunately I don’t think that will happen.
____________sponsored content ____________
Your home address, phone number, and family details are likely exposed on public websites—just a quick Google search away.
Stalkers, scammers, and creeps don’t need to hack you. They just need a people-search site.
I’ve been testing Incogni, and so far they’ve sent out 248 removal requests on my behalf. Of those, 205 have already been completed—saving me an estimated 153 hours of time. That’s time I didn’t have to spend emailing random websites or filling out endless opt-out forms.
Incogni removes your personal info from over 420 data broker sites automatically. But with their Unlimited plan, you can take it further. If you Google yourself and find your info on a site they don’t cover, just send them the link—and their privacy team will take it from there.
Stay safe. Stay private. Stop strangers from finding out where you live, work, or who you’re connected to. Because your personal safety should never be left to chance.
Be sure to use my coupon code PROFG at checkout to save.
____________sponsored content ____________
Newsletter Exclusive: How Palantir Turned Shareholders Into Superfans
Palantir isn’t just selling software anymore.
The defense-tech giant better known for its contracts with the Pentagon, is now pushing $119 tote bags and $99 “PLTR-TECH” gym shorts. The company launched new merch last week, the same week it rolled out an immigrant surveillance program with ICE. The merch store featured mock coding screens, and a note signed by CEO Alex Karp thanking buyers for “defending the West.” Apparel isn’t even their only offshoot — the Palantir Foundation runs a magazine, The Republic, commenting on technology, national security, and the role of the West in global affairs.
The goal of Palantir’s merch program is just to break even – making money isn’t the point. This is about symbolism and branding.
Palantir has done three merch drops since 2024, and the latest release features a T-shirt with a watercolor of Karp himself — wild hair, wire frame glasses, and the eccentric “mad scientist” look that’s become part of his persona. Palantir’s Head of Strategic Engagement, Eliano Younes, describes him as a “cultural icon,” arguing the merch taps into the same cultural moment as rapper or athlete tees. The campaign is part of a broader effort to frame Palantir as “THE lifestyle brand … the most pro-west, meritocratic, winning-obsessed brand on the face of the earth.”
Why does this matter? Because Palantir’s investor base is unusually retail-based. Roughly 50% of its shares are owned by individuals (versus 37% for the S&P 500 overall), and it’s consistently among the most-mentioned names on Reddit. Mentions of Palantir on investing forums across Reddit were up 375% in a 24-hour period following its earnings announcement last quarter, making it one of the most-discussed stocks among retail traders.
Palantir’s merchandise isn't just marketing swag; it's creating a tribal identity where investors see themselves as part of the “Palantir movement” rather than just holding shares in a data analytics company.
By fostering loyalty to Karp’s persona and the company’s narrative, Palantir creates what behavioral economists call “psychological ownership” — investors become emotionally invested beyond the numbers.
By turning superfans into shareholders — and shareholders into superfans — the company is cultivating a die-hard investor base eager to send the stock “to the moon,” regardless of fundamentals. So far, it’s working: Palantir trades at 131 times sales, the most expensive stock in the S&P 500 by a wide margin.
Palantir isn't alone in leaning on personality-driven branding. As Ed notes in his keynote People Are the New Brands, Nvidia leverages Jensen Huang and Tesla leverages Elon Musk — and both enjoy premium valuations. These companies have successfully created CEO-brand fusion where the leader becomes synonymous with the company's identity and future potential. This allows them to maintain premium valuations even when traditional metrics would suggest overvaluation (Tesla).
Want a playbook for winning the next generation of investors and customers? Book Ed Elson for a data-driven keynote on how to reach Gen Z.


Disney is going to be acquired or face an activist investor. It’s no longer too big — its market cap is $200 billion, the same as a decade ago while tech valuations have soared — and its single-class stock makes it vulnerable.

We’re honored to be nominated for a Signal Award, which highlights excellence in podcasting. Please vote for Prof G Markets here.

Prof G Media recommends: The Daily Upside, a daily source for sharp market insights and insider takes from Wall Street pros. Join over 1M savvy investors and subscribe for free today here.