There are always those pesky canaries in the coal mine. And along with them, there’s always the question: When will investors pay attention?
We’re referring to that event or events—such as a sharp drop in a significant player—that warn that a market bubble is getting closer to being unsustainable. Market tops, and subsequent crashes, don’t happen out of the blue, although it might seem like it at the time. There’s always something to tip you off. And recently in today’s market there have been warning signs ominously similar to those that preceded both the dot.com bust and the 2008 crash.
The tip-off in both those earlier two market bubbles was that as the end was approaching, it was the most leveraged stocks that were making the biggest gains—and that ultimately crashed the hardest. In the dot.com bubble, it was operational leverage, and a standout canary was Micron. In the case of the 2008 crash, it was financial leverage, with Bear Stearns an exemplar of excess. And these stocks were joined by others.
The current market bubble, of course, centers on AI with its supposedly unstoppable, civilization-changing prospects. The wild expectations for AI and for the companies that have anything to do with AI have driven the market’s rise, and they’ve been major drivers of GDP growth as well.
So what are today’s canaries? One of the biggest is former market leader Oracle. It’s one of the most financially leveraged companies you can find, borrowing wildly to support its AI ambitions. Oracle had aggressively pushed its way into AI through a deal it made with OpenAI, which was finalized in September 2025. Under that deal, Oracle is supposed to build—and build improbably fast—a slew of massive data centers for OpenAI. In return, over five years starting in 2027, OpenAI is supposed to reward Oracle by renting from it $300 billion in GPU clusters and computing capacity. Investors greeted the announcement of these plans with enthusiastic applause, and the stock instantly soared (making Oracle founder Larry Ellison briefly the richest man in the world).
Huge capital outlays, huge debt
Ellison, currently Oracle’s board chairman and chief technology officer, had hoped this ambitious plan would catapult Oracle to the top ranks of cloud computing providers, on a par with the likes of Amazon and Microsoft. But to realize these ambitions, Oracle has had to make huge capital outlays, which it has been financing with massive amounts of debt.
Was this a good bet? Notice our repeated use of the word “supposed” two paragraphs above. It turns out Oracle can’t build the data centers as fast as originally projected. In some cases, it has run into labor and material shortages. In one planned site in New Mexico, it was denied a natural gas pipeline it needed. The catch-22 is that if it doesn’t build the data centers, it won’t get paid by OpenAI. But if AI doesn’t have the data centers, it won’t be getting all the revenue the companies were counting on—though also true is that even if all the data centers were built, those revenues still might not match expectations.
Investors soured on Oracle this year, and its shares have plummeted sharply from their high of around $325 in the fall of 2025 to the current price of around $138, a drop of more than 60%. But while investors clearly have noted the problems with Oracle’s high spending and debt, it seems that they have yet to absorb what this fall in a major pillar of the AI buildout implies for the broader market.
A few weeks ago, the warning signs emitted from Oracle became even more insistent because of a downgrade of Oracle’s debt. On July 9, S&P Global lowered the credit rating of the company to BBB-, just one notch above junk bond status. S&P Global specifically linked the downgrade to the deal Oracle made with OpenAI and expressed doubt as to whether OpenAI would ever be able to meet its funding obligations to Oracle.
The degree to which the market and the economy now hinge on the rosy expectations for, and spending on, AI is, to say the least, alarming. AI-related companies account for roughly 30% of the valuation of the market and an estimated 60% of GDP growth. Everything rests on faith that the two major AI companies that the data centers are being built for—OpenAI and Anthropic, neither one of which has ever earned a penny—will become hugely profitable, with their AI models generating insatiable demand from enterprises that will use them to become more productive and profitable.
The Oracle debt downgrade says, whoa, not so fast.
Some of the best analyses of Oracle’s situation and of AI’s prospects in general come from Ed Zitron. Any of his YouTube videos or articles on Substack or elsewhere are highly informative and well-reasoned (not to mention bitingly funny). In a recent interview on the podcast The Tech Report, he lays out why the debt downgrade is so significant. A basic conclusion: OpenAI will never be able to pay Oracle the $300 billion it had committed to. Oracle’s entire AI venture will crater even as Oracle scrambles to borrow more money at higher rates. Oracle’s credit rating could be further downgraded—bringing it to junk bond territory. In the jargon of Wall Street, that would make it a “fallen angel.” If so, all the investment funds that hold Oracle bonds would be required to dump them, and Oracle’s borrowing costs would further balloon.
Oracle’s fate in and of itself isn’t all that significant. It’s not one of the major tech heavyweights in the market—on a recent day, for instance, it represented 0.55% of the S&P 500 compared to 7.50% for Nvidia and 4.29% for Microsoft. The point is that its sharp share decline and even more the debt downgrade are clear warning signs that the AI frenzy that has become the dominant driver of the stock market and the economy is getting closer to being exposed as wildly overoptimistic.
Competition from China
One thing that may increasingly weigh on AI companies’ prospects here is competition from Chinese AI models such as DeepSeek and Kimi K3. Unlike the proprietary “frontier” AI models of OpenAI and Anthropic, which charge money for access, the Chinese models are open source (or to be precise, “open mind”) and are available to anyone for next to nothing. And if you’re thinking that you get what you pay for, the Chinese models not only are free, they come within very close range of being as good as the U.S. models for almost all uses. It’s one more reason to think that the AI companies here will see considerably less demand than anticipated and that is needed to justify all the money being poured into their data centers.
Speaking of China, we want to mention a relevant opinion piece in Sunday’s The New York Times by the newspaper’s editorial board that seemed to us remarkably ill-informed and wrong-headed. Its basic argument was that the U.S. is currently beating China in the AI race; that it’s critical that we maintain our lead; that our edge comes from our having Nvidia’s more advanced chips; we therefore must continue to block China from getting these chips; and that without these chips, China will continue to lag us and we will remain triumphant.
What’s wrong with this argument? It ignores a reality that we’ve written about before: deprived of the advanced Nvidia chips, China has created a workaround known as logic folding that is letting it create the functional equivalent of those chips. The new Huawei phone scheduled for release next month will display this feat. And it’s a technological advance that applies as well to AI data centers in China, because it allows China to build data centers that are exponentially smaller and cheaper, and that consume far fewer resources, than those being built here. These advantages come on top of the fact that China has far greater renewable energy resources to power its centers and more land that’s remote from population centers on which to locate them.
There’s a lot more we could say about AI, but we’ll stop here. We think it should be clear why we worry that any day, doubts about whether AI really can deliver all that has been promised could reach a tipping point that would set off a major market correction with the potential to ripple through the economy. Meanwhile, the war in Iran is weakening the country in other ways, increasing inflation, further adding to our humongous government debt, and making our loss of hegemony increasingly evident—demonstrating the degree to which the order that has defined the world for so long is now eroding.
History shows that market manias can persist well beyond what’s reasonable, in fact, that’s the definition of a mania. There’s no way to predict when the end will come. Eventually, though, enough investors notice the canaries, and the party ends. When that happens, the majority of investors who had enthusiastically joined in find they can’t get out in time and get crushed in the stampede.
Guarding against that fate is why we have been stressing gold so insistently. Gold has been under pressure, and as we’ve always acknowledged, it could go down further. But it’s your best safeguard in today’s rapidly changing and chaotic world. We continue to believe that the U.S. economy is facing so many pressures that there’s no easy way out: We’re in for either a severe contraction and deflation or else—more likely—higher inflation. Either way, gold will protect you. In the case of deflation, it holds its value, which means you gain big in terms of spending power. And under inflation, we’d expect massive gains in the metal. These are not normal times, which is exactly when gold is the one thing you can count on.
Here are the benefits you unlock with a paid subscription:
- Full access to Turbulent Times Investor Newsletter (monthly edition)
- 60+ stock recommendations featured in our Core Investment Portfolios
- Weekly market updates via email, BUY/SELL alerts, and educational content
Our Substack blog, videos, and YouTube channel deliver compelling discussions on current events, geopolitics, macroeconomics, and educational content for investors. By raising awareness and empowering investors with knowledge, we enable them to make confident, well-informed investment decisions. Stephen Leeb, Ph.D., shares valuable wisdom and practical insights to help investors grasp market dynamics and geopolitical influences that present both risks and opportunities. Navigating the complex investment landscape can be overwhelming for the average retail investor, but our rigorous research on critical financial topics aims to equip investors with the tools to analyze independently and succeed in today’s market.
Follow the links below to visit Stephen Leeb’s official YouTube channel:
https://www.youtube.com/@StephenLeebPhD_TTI
Visit Stephen’s Interview “Archives” on YouTube!
