AIs Canary Starts Coughing: Lukes Peak Watch Ticks Higher as the AI Boom Faces a Potential Separation

The artificial intelligence sector, which has served as the primary engine for global equity markets over the last 24 months, is showing signs of structural strain as credit markets begin to price in significant risks. Last week, Elon Musk highlighted a growing sense of unease within the industry by reacting to the financial volatility surrounding CoreWeave Inc., a prominent "neocloud" provider that specializes in renting out massive clusters of Nvidia chips to AI developers. Within days of a series of financial disclosures, CoreWeave’s corporate bonds traded below 80 cents on the dollar, pushing the yield on its 2031 notes briefly above 16%. This shift into distressed-debt territory coincided with a 25% decline in the company’s perceived market value, marking a pivotal moment for an industry that has, until recently, enjoyed virtually unfettered access to capital.

The Financial Architecture of the AI Buildout

CoreWeave’s current predicament serves as a microcosm for the broader AI infrastructure boom. The company has funded its rapid expansion through aggressive borrowing, currently carrying approximately $52 billion in loans and lease obligations. While analysts anticipate CoreWeave will generate roughly $13 billion in revenue this year, the company’s planned capital expenditures are projected between $35 billion and $39 billion. This aggressive spending trajectory, coupled with a projected annual loss exceeding $2 billion, has raised concerns among institutional lenders regarding the company’s long-term ability to service its debt.

The volatility was exacerbated by recent macroeconomic data. A higher-than-expected inflation report last Wednesday increased the probability of further interest rate hikes by the Federal Reserve, raising the cost of future borrowing. By Thursday, the cost of credit default swaps (CDS)—the insurance used to protect against a company failing to pay its debts—for CoreWeave reached record highs. Market data indicated that protecting $1 million of the company’s bonds now costs approximately $119,500 annually. Technology analysts, including Luke Lango, editor of Innovation Investor, note that the bond market often serves as a leading indicator of trouble before the equities market fully reacts.

Monitoring the Peak: The 13-Signal Checklist

To navigate this period of heightened volatility, market strategists have developed tools to identify when a sector-wide boom transitions into a bust. Lango’s "Peak Watch Checklist" tracks 13 specific warning signals categorized into three distinct phases: Froth, Cracks, and Confirmation. As of the latest reading, seven of these 13 signals have turned red, indicating that the market has moved beyond mere speculation into a period of systemic vulnerability.

The "Froth" stage is currently deemed fully active. This phase is characterized by Big Tech firms borrowing heavily to fund capital expenditures, suppliers providing financing to their own customers to maintain demand, and record-breaking levels of stock issuance. Recent market movements have seen credit concerns spread from specialized players like CoreWeave to broader infrastructure providers. Furthermore, the recent sell-off in Taiwan Semiconductor Manufacturing Company (TSMC) shares, despite strong earnings results, suggests a "sell the news" sentiment that often precedes a market correction.

Despite these warnings, vacancy rates in North American data centers remain at a record low of 1.4%, and high-end AI chips continue to sell out immediately upon release. Long-term projections suggest a 60% probability that AI-generated revenue will eventually justify current spending levels, though full payback for many infrastructure projects may not be realized until the late 2030s. This creates a significant "duration risk" for investors who may not have the stomach for a decade-long financing cycle.

The Separation: Builders vs. Appliers

As the AI boom matures, a divergence is emerging between two classes of companies: the "Builders" and the "Appliers." This phenomenon, termed "The Separation" by global macro expert Eric Fry, echoes the market dynamics of the late 1990s and early 2000s. During the dot-com era, companies like Corning, which manufactured the fiber-optic cables for the internet’s infrastructure, were initially the most sought-after stocks. However, as overcapacity led to a collapse in fiber prices, Corning’s stock plummeted by 87%. Conversely, companies like Humana, which were not tech firms but used the new internet infrastructure to streamline operations and improve efficiency, saw their valuations increase by thousands of percentage points over the subsequent decades.

In the current landscape, "Builders" include the likes of Microsoft, Meta, and Micron—companies spending or supplying hundreds of billions of dollars for AI infrastructure. The "Appliers" are companies in diverse sectors, such as healthcare, energy, and consumer staples, that are utilizing increasingly affordable AI computing power to optimize their business models. Analysts warn that the Builders are currently caught in a "vise" formed by two competing economic forces: the Revenue Jaw and the Expense Jaw.

The Revenue Jaw: The Deflation of Computing Power

The Revenue Jaw refers to the rapidly declining cost of AI outputs. As AI models become more efficient and chips more powerful, the price of "tokens"—the basic units of AI processing—has plummeted. In September, token prices hit a record low, falling more than 50% from their summer peak. While this is beneficial for companies using AI, it threatens the profit margins of companies that built expensive data centers based on yesterday’s higher pricing models. If GPU (Graphics Processing Unit) rental prices begin a sustained decline to match token prices, the revenue models for many infrastructure "Builders" could collapse.

The Expense Jaw: Rising Costs and Regulatory Pushback

Simultaneously, the Expense Jaw is tightening. Data centers are becoming increasingly expensive to construct and operate. Beyond the cost of capital, there is significant political and social pushback against the proliferation of these facilities. In September, the U.S. House of Representatives voted 417-3 in favor of legislation requiring data centers to pay the full cost of the power grid upgrades they necessitate, rather than passing those costs on to local taxpayers.

Furthermore, public sentiment is shifting. A recent Gallup poll found that 71% of Americans oppose the construction of new data centers in their local communities, citing concerns over rising electricity bills and environmental impact. At least 20 candidates in upcoming gubernatorial races have proposed moratoriums on data center development. These regulatory hurdles increase the time and cost required to bring new capacity online, further squeezing the margins of infrastructure providers.

Market Rotation and the "DeepSeek" Effect

Recent market behavior suggests that capital is already beginning to migrate in anticipation of this "Separation." On January 27, 2025, Nvidia lost a staggering $589 billion in market value in a single day following reports that China’s DeepSeek had developed highly efficient AI models using significantly cheaper hardware. Despite this massive sell-off in the tech sector, the Dow Jones Industrial Average finished the day higher, suggesting that investors were not exiting the market, but rather rotating into different sectors.

This trend has continued into the current quarter. The iShares Semiconductor ETF (SOXX) has entered bear-market territory, falling 25% from its June peak. During the same period, defensive sectors have seen renewed interest. The Consumer Staples Select Sector SPDR ETF (XLP) recently posted its best weekly performance since early 2024, while the Energy Select Sector SPDR ETF (XLE) has also seen significant inflows. This rotation indicates that the market is beginning to prioritize companies with strong cash flows and lower capital intensity over high-growth, high-debt tech firms.

The Two-Week Gauntlet

The financial community is now bracing for a two-week period that could determine the trajectory of the AI trade for the remainder of the year. This "gauntlet" includes earnings reports from Meta, Alphabet, Microsoft, and Amazon. Analysts will be scrutinizing these reports not just for revenue growth, but for any signs that financing stress is altering capital expenditure plans. If these tech giants signal a reduction in chip orders or the cancellation of data center projects, the "Cracks" signal on the Peak Watch Checklist could move toward "Confirmation."

Furthermore, the Federal Reserve’s upcoming meeting and the U.S. midterm elections on November 3 add layers of macroeconomic and political uncertainty. Regardless of the election outcome, the bipartisan pushback against data center tax breaks and energy consumption suggests that the regulatory environment for "Builders" will remain challenging.

Conclusion: Navigating the New AI Landscape

The current volatility in the AI sector does not suggest the technology is a fad, but rather that the "infrastructure phase" of the boom is reaching a point of saturation. History suggests that while the companies building the "plumbing" of a new technology often face the most significant risks during a market correction, the companies that successfully apply that technology to existing industries often provide the most sustainable long-term returns.

As the "Separation" accelerates, the distinction between companies with pricing power and those with heavy debt burdens will become the primary driver of portfolio performance. For investors, the coming weeks represent a critical window to assess exposure to the "Builders" and identify the "Appliers" who are quietly leveraging AI to transform traditional business models. The AI boom is not ending, but its rewards are beginning to shift toward those who use the tools, rather than those who merely provide the parts.