Is the AI Bubble About to Burst? 13 Signals We’re Watching
The financial markets currently stand at a critical juncture, characterized by unprecedented capital expenditure in artificial intelligence and growing investor scrutiny regarding the long-term profitability of these investments. While the current trajectory of AI remains bullish in terms of technological adoption and infrastructure buildout, historical precedents suggest that every significant technological boom eventually reaches a point of exhaustion. To navigate this landscape, analysts have developed the Peak Watch Checklist, a systematic framework consisting of 13 specific signals designed to identify when the AI boom transitions from a growth phase into a speculative bubble nearing its collapse. Currently, the checklist indicates that seven of these 13 signals are "lit," suggesting the market has entered a late-cycle phase, though a fundamental peak has not yet been confirmed.
Historical Context: The Lessons of 1873 and 2000
To understand the current risks inherent in the AI sector, one must look back to the collapse of Jay Cooke & Co. on September 18, 1873. Jay Cooke was arguably the most influential banker in 19th-century America, having gained national trust by financing the Union’s efforts during the Civil War through the sale of government bonds. His downfall was precipitated by his involvement in the Northern Pacific Railway. When the public appetite for railroad bonds waned, Cooke’s firm began quietly absorbing the debt itself, essentially financing its own customer to maintain the appearance of growth.
The resulting panic was so severe that the New York Stock Exchange was forced to suspend trading for ten days. In retrospect, the warning signs—tightening liquidity and the decoupling of stock performance from positive news—had been visible for over a year. Similar patterns were observed during the dot-com bubble of 2000, where exuberant valuations eventually collided with the reality of slow revenue realization. The AI buildout, while fundamentally different in its technological utility, is subject to the same economic gravity that governed the railroads and the early internet.
The Peak Watch Framework: A Three-Stage Methodology
The Peak Watch Checklist categorizes the lifecycle of a market peak into three distinct stages. Each signal is assigned a status: Clear (green), Watch (yellow), Flashing (orange), or Triggered (red).
Stage 1: Market Froth
This stage is characterized by the abundance of "cheap" money and circular financing patterns. It represents the psychological peak of the boom where enthusiasm overrides traditional fiscal discipline. Currently, this stage shows significant activity:
- Debt-Funded Capex (Triggered): Major technology firms, colloquially known as the "hyperscalers," have shifted from spending cash reserves to increasing debt loads to fund AI infrastructure. Alphabet, Amazon, and Meta have recently reported negative free cash flow as a direct result of these investments. By mid-2025, bond issuance from these entities had already surpassed the total for the previous year.
- Vendor Financing (Flashing): This occurs when a supplier provides the capital for a customer to buy its own products. Notable examples include Broadcom’s $42 billion financing arrangement for Anthropic and reports of a $50 billion credit facility for OpenAI.
- Record Stock Issuance (Flashing): Companies often issue stock when they believe their valuation is at or near a peak. Meta’s $35 billion share sale and Alphabet’s $85 billion issuance are indicative of this trend.
Stage 2: Structural Cracks
The second stage involves the formation of the market top. During this phase, even positive earnings reports fail to drive stock prices higher, and the cost of capital begins to rise.
- "Sell the News" Sentiment (Flashing): Alphabet recently saw its stock price decline by more than 12% following an earnings report that featured a 103% growth in its Cloud business. Similarly, Meta experienced a 7% slide despite beating analyst expectations. This suggests that "perfection" is already priced in, and investors are looking for reasons to exit.
- Credit Spread Widening (Triggered): The bond market often serves as a leading indicator for equity markets. AI-centric firms are facing higher borrowing costs; CoreWeave’s bonds are currently yielding between 15% and 16%, while Volta recently secured a $5 billion loan at an interest rate exceeding 11%.
- Monetary Policy (Flashing): Historically, booms are ended by central bank intervention rather than a lack of demand. The Federal Reserve’s recent rate hike in September, combined with a 90% probability of a subsequent hike in December, indicates a tightening environment that could stifle speculative growth.
Stage 3: Fundamental Confirmation
The final stage is the actual "rolling over" of the industry’s fundamentals. This is where the peak is confirmed through tangible data regarding supply and demand.
- Chip Demand (Clear): Currently, this remains the strongest pillar of the AI boom. Taiwan Semiconductor Manufacturing Company (TSMC) and ASML have both maintained or raised their outlooks, citing that demand for AI-specific chips continues to outpace supply.
- Project Pipelines (Watch): While most projects remain on track, political and logistical hurdles are emerging. Governors in Georgia and Texas have recently campaigned on data-center moratoriums due to power grid concerns. To date, only a few major projects, such as Firmus’s 1.6-GW Australian facility, have faced outright cancellation.
Timeline of Upcoming Market Catalysts
The remainder of the 2025-2026 fiscal cycle will be defined by several high-stakes events that will determine the Peak Watch Score’s next movement.
- Nvidia Earnings Report (November 17): As the primary provider of AI hardware, Nvidia’s performance is the ultimate litmus test for the "chip orders digest" and "good news gets sold" signals. Any guidance suggesting a plateau in demand could trigger multiple Stage 3 signals simultaneously.
- Anthropic Initial Public Offering (IPO): Following the withdrawal of the Firmus IPO due to weak demand, the Anthropic listing will serve as a definitive gauge of investor appetite for AI infrastructure at high valuations.
- Federal Reserve Meeting (December 8-9): With the October jobs report showing 165,000 jobs added—significantly higher than the 59,000 expected—the Fed is expected to remain hawkish. A second consecutive rate hike would move the "Fed Tightens" signal from Flashing to Triggered.
Analytical Perspective: The Payback Gap
A critical question facing the industry is whether the current level of capital expenditure will yield a proportional return on investment. Current financial modeling suggests a significant "payback gap." In a base-case scenario, the revenue generated by AI applications may not clear the required-return threshold until approximately 2035, with a full payback of initial infrastructure costs not occurring until 2039.
This decade-long window creates a period of vulnerability. While the long-term utility of AI may be validated, the interim period is likely to be marked by "financing scares" and sharp market corrections as investors lose patience with the slow pace of monetization. This distinguishes a "late-cycle" market from a "collapsed" market. Historically, some of the most significant gains in a technological boom occur during the final innings, provided investors can distinguish between a temporary shakeout and a permanent top.
Implications for the Global Economy
The transition of the AI boom into a late-cycle phase has broader implications for global liquidity and technological development. If the "Froth" signals continue to trigger without a corresponding increase in "Confirmation" signals, the risk of a "hard landing" for the tech sector increases.
Conversely, if the "Confirmation" signals—such as chip orders and data center occupancy—remain "Clear" while "Cracks" in financing are repaired, the boom could extend into a more sustainable, albeit slower, growth phase. The current Peak Watch Score of 7 out of 13 suggests that while the "music is still playing," the exits are becoming more crowded. For institutional and retail investors alike, the focus has shifted from pure growth to "eyes on the scoreboard," monitoring whether the fundamental data can continue to support the increasingly expensive cost of AI development.
The AI trade has officially moved beyond its "clean" early innings. The market is now witnessing a separation between strong borrowers and weak ones, and between companies with actual AI revenue and those merely riding the wave of sentiment. As the checklist updates weekly, the financial community remains focused on the "tripwires" that would turn the remaining green signals to red, marking the definitive end of the first great AI expansion.