Like clockwork, similar scenarios unfold every six months. "AI bubble" theories consistently emerge on schedule, triggering brief market panic before being quickly overwhelmed by renewed waves of enthusiasm.

From Goldman Sachs questioning commercial returns, to China launching highly cost-effective models, to Oracle and OpenAI announcing a market-shaking $300 billion "future contract," AI skepticism and euphoria alternate in endless cycles.

However, analysis suggests that beneath this cyclical debate, a deeper structural risk is emerging: the AI infrastructure race is evolving from a marathon supported by tech giants' internal cash flows into a debt-dependent "arms race."

When a $1.5 trillion funding gap needs to be filled by an already strained private credit market, one must ask: how far away is that inevitable "wolf"?

**AI Bubble Theories Right on Schedule**

The first wave of large-scale concerns emerged in June 2024. Goldman Sachs released a report directly questioning whether generative AI represents a "too much investment, too little return" capital black hole - potentially a bottomless pit that may never deliver long-term positive returns for investors. This skepticism dropped a bombshell in the tech world.

However, six months later, after burning through another $100 billion to "perfect" the planet's most expensive chatbot, clear profitability models still seem absent in the United States. Instead, China launched the highly acclaimed DeepSeek large model, which is not only open-source but significantly cheaper than American counterparts, requiring far less expensive equipment than the latest NVIDIA super graphics cards.

Meanwhile, reports suggest Microsoft, Google, and Meta are quietly reducing capital expenditures. These factors combined triggered another round of AI concept stock sell-offs, lasting from late January through April.

History appears to repeat itself. This inevitably recalls the first internet bubble period, when once-prominent companies ultimately couldn't escape bankruptcy.

**"Infinite Money Glitch": When Financing Shifts from Cash to Debt**

Fast forward to September 2025, the AI bubble is expanding at full speed, single-handedly pushing stock markets to their highest valuation levels since the internet bubble...

However, Oracle shattered this celebration's tranquility on September 10 with an extremely bold move. According to reports at the time, it announced a five-year, $300 billion cloud computing agreement with OpenAI. This was viewed as one of history's largest "vendor financing" deals.

More impactful was Oracle's simultaneous reminder to everyone - it actually lacks sufficient internal cash to fund this spending spree expected to continue into the 2030s. So where does the money come from? Borrowing.

JPMorgan analyst Michael Cembalest, in his latest "Market Watch" report, eloquently described this AI circular economy that many peers call an "infinite money glitch."

He used a simple circular diagram to explain this phenomenon: AI companies promise future massive payments to cloud service providers → Cloud providers use this story to borrow money for infrastructure construction → Infrastructure is then leased to AI companies.

Cembalest noted that since ChatGPT's launch in November 2022, AI-related stocks have contributed 75% of S&P 500 returns, 80% of earnings growth, and 90% of capital expenditure growth. Data center power consumption is driving up electricity prices - for example, in the PJM region, 70% of last year's electricity price increases can be attributed to data center demand.

Oracle and OpenAI's deal perfectly exemplifies this "glitch." Doug O'Laughlin commented:

Oracle simply cannot pay for all this with cash flow; they must issue stock or take on debt to realize their ambitions... A stable oligopoly structure is fracturing... What was originally a disciplined, cash flow-funded competition may now become a debt-driven arms race.

**$1.5 Trillion Funding Gap - Can Private Credit Fill It?**

Oracle's case reveals a deeper issue: AI infrastructure construction costs are spiraling out of control, far exceeding tech giants' internal cash generation capabilities. A Morgan Stanley report paints this shocking picture: global data center-related spending is projected to reach approximately $2.9 trillion by 2028.

The report indicates that while large tech companies' internal cash flows remain the primary funding source, after accounting for shareholder returns and other factors, they can self-finance at most around $1.4 trillion. This means the market faces a massive $1.5 trillion funding gap.

Morgan Stanley believes credit markets will play an increasingly important role in bridging this gap.

Among all credit channels, private credit is highly anticipated. The bank expects that among various capital sources filling the gap, private credit (especially asset financing) will contribute approximately $800 billion, becoming the most important external funding source. Consulting firm Bain subsequently reached similar conclusions.

AI's future appears deeply tied to private credit's purse strings.

**Private Credit: AI's "Savior" or "Achilles' Heel"?**

However, betting AI's future on private credit may be a dangerous gamble. Just as markets expect it to provide AI "blood transfusions," the industry's own health has shown warning signs.

Market data shows that BXSL, the private credit fund under Blackstone Group (one of the world's largest private credit managers), has fallen to 2025 lows, performing far worse than the S&P 500. Another industry giant, Blue Owl, faces similarly precarious stock performance. Blue Owl has already deeply engaged in AI sector financing activities.

These private credit titans' troubles extend far beyond providing data center funding. They're heavily exposed to the weakest links in the U.S. economy - consumers, especially low-income groups where non-performing loans (NPLs) are soaring in the "buy now, pay later" (BNPL) sector.

The private industry is "sitting on $5 trillion of existential fear." If this industry, viewed as AI's financial backbone, faces its own troubles, where will the promised $800 billion for AI come from?

**"Bubble" Within the Bubble: When No One Discusses Bubbles Anymore**

As financial structural risks become increasingly apparent, public discussion about AI bubbles is cooling. Deutsche Bank analyst Adrian Cox points out that global "AI bubble" internet search volume has declined 85% from its August 2025 peak. In other words, "the bubble of discussing AI bubbles" has itself burst.

But this doesn't mean the all-clear has sounded. History shows asset bubble evolution isn't a linear process. In the five years before the 2000 dot-com bubble burst, the NASDAQ experienced seven corrections exceeding 10%.

More importantly, in November 1998, when investment manager Michael Murphy warned "this is a serious bubble," the NASDAQ was still below 2,000 points. It continued doubling over the next 16 months, breaking through 5,000 before finally collapsing.

**When Will the Wolf Actually Come?**

After crying "wolf" every six months, markets seem fatigued. Oracle's massive deal reveals dangerous signals of AI prosperity shifting from "doing" to "borrowing," while the expected financier - private credit - is itself mired in difficulties.

This debt and dream-driven celebration appears even more fragile under hard constraints like power grid infrastructure.

Perhaps when everyone stops discussing bubbles, the wolf quietly approaches the door. Or, as that old market saying goes: markets can remain irrational longer than you can remain solvent.

So are we living through history's largest bubble? When will it burst?

The honest answer is: nobody knows. Just as NVIDIA's stock hits new highs and market cap soars to an astonishing $4.5 trillion, markets still buy into the AI narrative.