Getting Real

The market's balancing act... Keep watching the labor market... More parallels with the dot-com era... Hundreds of billions in spending... Winners can keep winning... A few reasons for pause... You can see it coming...

Good news, bad news...

We've discussed before how poor news in the real world – like more people losing jobs or finding it harder to get one – can be perversely "good" news in the market. Economic pain strengthens the idea of a "rescue" coming via easier policy from the Federal Reserve.

Well, the opposite can also be true. Good real news can be taken as "bad" in the market, especially when expectations are for the Fed to help the economy with a series of interest rate cuts ahead.

Today, we saw an example...

A surprise in the jobs market...

This morning, the Bureau of Labor Statistics' weekly jobless claims bucked the recent trend of softening labor-market data. In the week ending September 20, 218,000 people filed for first-time unemployment benefits.

That was below the prior week's level of 232,000 and better than Wall Street expectations – and it marked the lowest level since July. And it's a huge change from just a few weeks ago. As we wrote in the September 11 Digest...

When you exclude the COVID-19 pandemic, jobless claims sit at the highest level since September 2017.

This is just the latest in bad labor market data. In the past week alone, we've covered weak job openings, another bad payroll report, and the huge downward revision in job creation over the past 12 months.

But just two weeks later, initial jobless claims are now at a two-month low. And continuing claims – which represent folks that have been filing for unemployment benefits for multiple weeks – are steady, too.

Continuing claims, which run on a one-week lag to initial claims, came in at 1.93 million for the week ending September 13. Continuing claims have been bouncing around between 1.9 million and 1.96 million since about May.

So while that's still at the highest level since 2018 (excluding the pandemic), it doesn't represent the huge spike higher that we may expect during a softening labor market.

In response today, the futures market now gives slightly lower odds of further rate cuts this year. And bond yields rose (again), while simultaneously, the U.S. stock indexes fell for a third straight day.

This will be behavior to keep an eye on, at least over the next few weeks. The widespread expectations for rate cuts are a tailwind for the market. We want to see if they remain or they soften.

It's not a trend reversal – yet. Labor-market risks remain...

In a speech on the economic outlook earlier this week, Fed Chair Jerome Powell repeated his stance that risks have increased. In his prepared statement, he repeated what he said during last week's post-meeting press conference...

Payroll job gains slowed sharply over the summer months, as employers added an average of just 29,000 per month over the past three months. The recent pace of job creation appears to be running below the "breakeven" rate needed to hold the unemployment rate constant.

Put another way, unless hiring picks up, Powell expects the unemployment rate to move even higher. And there are reasons to believe that it might...

As the Kobeissi Letter shared on X this week, an index measuring job postings on the hiring site Indeed fell 8% year over year in the week ending September 12. And it now sits at the lowest level since February 2021...

The index goes back to February 2020, just before the start of the pandemic. An index value of 100 represents the number of job postings on February 1, 2020, before the world was turned upside down. (We've started our chart in late 2020 to leave off the early days of the COVID crash and rapid reversal.)

In the week ending September 12, the index read 103. So there are only about 3% more postings on Indeed for jobs than there were during pre-pandemic times five years ago. And the index is down 35% from its 2022 highs.

This trend in hiring plans doesn't suggest a strong outlook for the labor market.

Here's our point... One week of lower jobless claims is not going to undo all of the red flags we're seeing in the labor market. Hiring is down, and the average length of unemployment is rising (now sitting at a three-year high of 24.5 weeks).

And that's what the Fed is seeing as it plans two more rate cuts this year. The labor market is going to remain the key focus for the Fed, as long as inflation doesn't begin to take off.

Speaking of inflation...

The next round of important inflation data comes out Friday morning, when the Bureau of Labor Statistics releases personal consumption expenditures ("PCE") data for August.

The Cleveland Fed expects PCE and core PCE (excluding food and energy) to come in at 2.8% and 3% on a year-over-year basis, respectively.

For now, that's still a level where the Fed is comfortable cutting rates. But if rate cuts and perhaps tariffs push prices even higher, or the labor market starts to rebound, the Fed will have to rethink its policy outlook.

And that shift could dampen the market mood.

We'll keep an eye out for more signs like we saw today: All but the energy sector of the S&P 500 Index were lower. And even gold – which has been moving steadily higher in part because of rate-cut expectations – rose by a smaller-than-usual 0.3%. Bitcoin fell to less than $110,000.

So at least some people are concerned with what we're talking about... But over the longer term, the market is still full of optimism today.

The analogy that never stops...

We keep writing about comparisons of today's AI-driven bull market and the buildup to the dot-com bubble nearly three decades ago.

Maybe you're tired of it... Or perhaps you feel we're too negative (though the comparison isn't all negative, as we'll explain today). But the similarities are just too striking to ignore.

Check out this latest comparison that our colleague and Stansberry's Investment Advisory lead editor Whitney Tilson shared in his free daily newsletter today...

Here's an interesting graph courtesy of InvestorPlace, highlighting the similar trajectories of tech stocks from 1995 to 1999 compared with the past three years:

It's easy to create overlapping historical charts that mean absolutely nothing, so I don't want to overweigh this one.

But it somewhat reinforces the possibility that in the next year or two, we could see a melt up to a bubble like what occurred in 1999 and early 2000, followed by a bust.

There's that phrase we wrote about earlier this week: Melt Up. As you can see in the chart above, even during the lead-up to one of the biggest market bubbles of all time, the Nasdaq's rise was not uninterrupted... but the overall rise lasted for years.

Along the way, though, the index faced frequent pullbacks and a major drop in 1998 before the heights of mania in 2000.

Today, in the really short term, we could see reason for pause in the ongoing bull market. The labor market remains uncertain, and with it, investor confidence in rate cuts. But pause or not, it's still a bull market.

The AI trend has powered stock market growth for the past few years – and made AI-chip maker Nvidia (NVDA) 7% of the entire market-cap-weighted S&P 500. And it doesn't appear to be slowing down...

They're still willing to spend...

Four "hyperscalers" – Microsoft (MSFT), Meta Platforms (META), Alphabet (GOOGL), and Amazon (AMZN) – are planning to pour hundreds of billions into AI this year alone. Stansberry Research senior analyst Brett Eversole covered this in a piece for our Stock Market Trends site earlier this week. And as Brett continued...

If we build a holistic look at AI spending, the numbers get even larger. Here's what estimates look like over the next few years...

This is a mountain of spending... Spending that never would have happened if AI hadn't hit the scene.

These estimates show total spending growing well past $1 trillion. And it seems safe to say we'll be looking at a multitrillion-dollar spending boom by the end of the decade.

There are two ways to look at that... as something to fear... or as a major opportunity. I suggest you view it as the latter...

Now, there are also signs that should give pause...

Big Tech CEOs are more than content to spend hundreds of billions of dollars with no possible defined expectation for a return on investment. They get the satisfaction of "not missing out on AI" in whatever form it ultimately sticks... but no guarantee they'll turn a profit from this costly venture.

And Nvidia is becoming an ever-growing financier for the AI buildout. See this week's announcement from Nvidia to invest up to $100 billion in OpenAI, the firm known for ChatGPT, to build data centers... using Nvidia chips and technology.

And plenty of promises being made leave room for questions... like OpenAI saying it will spend $300 billion on infrastructure from Oracle (ORCL), while OpenAI CEO Sam Altman also just told the Wall Street Journal this week...

I don't think we've figured out yet the final form of what financing for compute looks like.

But I assume, like in many other technological revolutions, figuring out the right answer to that will unlock a huge amount of value delivered to society.

And here's Meta CEO Mark Zuckerberg on a podcast last week...

If we end up misspending a couple of hundred billion dollars, I think that that is going to be very unfortunate obviously. But what I'd say is I actually think the risk is higher on the other side.

If you build too slowly and then superintelligence is possible in three years, but you built it out assuming it would be there in five years, then you're just out of position on what I think is going to be the most important technology that enables the most new products and innovation and value creation and history.

We've shared this warning before...

Yes, there will be products and uses for AI that most people can't imagine right now. And they will require infrastructure (both data centers and energy to power them) that does not exist yet. And there will be money made in the right investments. That's the exciting part.

But surely, there will be wasted investment, too, some of which will look obvious in hindsight... commitments never met, products that never find a market, and dreams not realized. It's time to start talking about this more.

We wrote about this a few weeks ago regarding Oracle's 40% one-day stock-price pop. The company issued new earnings projections based on massive contracts it may never fulfill, and feverish investors treated the claim like money in the bank. (ORCL shares are down about 14% from that peak.)

The Nvidia-OpenAI deal, as announced Monday, similarly is a "letter of intent for a landmark strategic partnership to deploy at least 10 gigawatts of Nvidia systems for OpenAI's next-generation AI infrastructure to train and run its next generation of models on the path to deploying superintelligence" with the "first phase targeted to come online in the second half of 2026."

Here's former Stansberry Investor Hour guest Larry McDonald, editor of The Bear Traps Report and a former Lehman Brothers trader, with a sobering dose of reality about the real energy needs for the project...

Larry went on to point out that building a traditional nuclear power plant in the U.S. typically takes around seven to 10 years (and small modular reactors two to seven years). And while the Trump administration could cut the timeline down, he says that "it's time to GET REAL."

We like 'real' as much as anyone, but we also acknowledge, as John Maynard Keynes famously said...

"Markets can remain irrational longer than you can remain solvent."

For now, it seems enough people are content to ignore some rationality and ride the emotional buzz and optimism about AI higher. And so, this boom can go on for longer than you or I think it "should" – and AI winners can keep winning.

As Brett Eversole also wrote in his Stock Market Trends piece about the Nvidia-OpenAI deal and the AI boom...

Seeing major deals between suppliers and customers might feel a tad incestuous. But as we'll see, it's just another proof point that the huge amount of AI spending we've seen isn't over yet. Instead, we're just hitting the tip of the iceberg...

I'm not saying this won't end badly. Most technology booms follow the same trajectory...

An exciting new opportunity leads to a massive infrastructure buildout. That fuels a speculative mania. But we always end up overbuilding. And that leads to a crash.

There's a good chance the AI boom follows a similar path. But we're nowhere near the end...

The Economist notes that Britain experienced a railway boom in the 1840s. From 1844 through 1847, investment in railways grew from 5% of GDP all the way to 13%. That was an unsustainable level, and it led to a painful bust.

We're not near those unsustainable levels right now. AI spending in total is around 3% to 4% of GDP over the past four years. Said another way, investment needs to grow fivefold to hit the excesses we saw in Britain's railway boom.

That probably will happen by the end of the decade. But it won't happen overnight. And in the meantime, this massive amount of AI spending will fuel a continued boom in the U.S. stock market.

Zuckerberg was asked in this podcast if AI might be in a "bubble" already. He replied that "it's quite possible" based on historical infrastructure buildouts of railroads or fiber networks during the dot-com boom and bubble.

And yet, his company is still firing a bazooka of cash at AI projects.

So it goes.

We wouldn't be short AI right now. By all means, if you find worthy investments in the trend, you can enjoy the ride higher.

Just know the risks and be prepared to exit when, if history is any indication, mass hysteria ensues and a lot of people get in trouble.

And, as always, hedge your speculations with a well-diversified portfolio that includes shares of great companies outside the world of AI... plus "hard assets" like gold.

And it's a good idea to have cash to put to work. When booms inevitably bust, they present great long-term buying opportunities that come around about only as often as world-changing technologies.

2025 Stansberry Conference & Alliance Meeting
In-Person Registration Closes Tomorrow!

If you're interested in attending our annual Stansberry Research Conference, you have just one day left to secure your ticket to our biggest event of the year... In-person ticket registration ends this Friday, September 26.

This year's Las Vegas conference is sure to impress, with an incredible speaker lineup (including entrepreneur Peter Diamandis, market commentator Josh Brown, and journalist Kara Swisher) and your favorite Stansberry Research editors and friends in attendance.

Dozens of ideas and stock recommendations are shared onstage...

You can see all the details and register here, and we hope to see you there.

New 52-week highs (as of 9/24/25): Altius Minerals (ALS.TO), Antero Midstream (AM), Alpha Architect 1-3 Month Box Fund (BOXX), Cencora (COR), Lynas Rare Earths (LYSDY), Medtronic (MDT), and Valero Energy (VLO).

A quiet mailbag today... What's on your mind? Do you have thoughts on our comparisons about the AI boom and the dot-com days? As always, send your comments and questions to feedback@stansberryresearch.com.

All the best,

Corey McLaughlin and Nick Koziol
Baltimore, Maryland
September 25, 2025