This round of AI stock crash: A global stock market disaster triggered by a storm

A seemingly ordinary rainstorm triggered a market shock in AI infrastructure stocks worth hundreds of billions of dollars.

Data center service provider CoreWeave saw its market value evaporate by $33 billion in six weeks, a decline of over 60%. Meanwhile, Broadcom and Oracle both plummeted by more than 17% within three trading days, highlighting that market concerns about the AI bubble are spreading from delayed capital expenditures to the entire supply chain.

This crisis began at a construction site in Denton, Texas. According to a report by The Wall Street Journal on Monday, strong summer rains and high winds caused a delay of about 60 days for a large AI data center cluster from CoreWeave, as contractors were unable to pour concrete as scheduled. The cluster was planned to install about 260 megawatts of computing power and lease it to OpenAI, but the completion date has been postponed by several months.

CoreWeave CEO Michael Intrator's chaotic remarks during the November earnings call further exacerbated investor panic. He initially attempted to downplay the impact, claiming that "only one data center" had issues, but was promptly corrected by the CFO, who stated that it was actually "one data center supplier" facing delays, implying a broader problem. Following this chaotic statement, CoreWeave's stock price plummeted by 16.3% in a single day.

This incident exposed the systemic risks facing the entire AI infrastructure industry: the gap between the frenzied pace of construction and actual delivery capabilities is widening, while valuations in the hundreds of billions of dollars have already priced in optimistic expectations.

Chain Reaction Triggered by Rainstorm

According to informed sources, the unusually strong summer rains and high winds in northern Texas caused about 60 days of delays at CoreWeave's construction site in Denton, north of Dallas. The construction of this data center cluster is being carried out by Core Scientific, headquartered in Austin, which was once a cryptocurrency mining company and has now become one of CoreWeave's largest landlords.

In addition to weather factors, some data centers being built by CoreWeave and its partners in Texas and other regions have also faced additional delays due to design modifications. These delays have postponed the delivery of the large computing cluster that CoreWeave planned to lease to OpenAI by several months.

CoreWeave's business model relies on high-interest debt to purchase thousands of advanced AI chips from Nvidia, installing them in server racks leased from third-party landlords, and then subleasing the chip usage rights to AI companies. This highly leveraged model appears particularly fragile in the face of construction delays.

CEO's Contradictory Statements Create Panic

Intrator attempted to calm investor concerns about the delays during the earnings call on November 10. He initially stated that "one data center has issues" and claimed, "this data center will catch up, and then we will move forward." However, CFO Nitin Agrawal quickly corrected him, stating that the delays were concentrated on "one data center supplier," rather than a single data center, indicating that the problem is more widespread. In the same conference call, Intrator described construction delays as "systemic challenges" and stated that this "has left our customers very frustrated." The company is trying to diversify its data center construction suppliers to mitigate the impact of inevitable delays. This contradictory statement has unsettled investors.

The next day, during an interview with CNBC's Jim Cramer, Intrator repeated the phrase "one data center" and only corrected himself after the host prompted him. On that day, CoreWeave's stock price fell from $105.61 on the day of the earnings report to $88.39, a decline of 16.3%. The stock price continued to decline until December.

High Debt and Profitability Issues

D.A. Davidson analyst Gil Luria stated that CoreWeave has "the ugliest balance sheet in the tech industry to date." CoreWeave's revenue doubled year-on-year to nearly $1.4 billion in the most recent quarter, but the company is still not profitable, posting a loss of $110 million in the latest quarter.

Luria pointed out that CoreWeave's operating profit margin of about 4% is less than half of the interest expenses on most of the debt the company uses to deploy customer computing capacity, making it difficult to see how profits will be generated in the future. "Bulls believe the company will achieve profitability through scaling, and many companies start with low margins, but this is already a company with scale, and there is no scaling happening here."

Last week, CoreWeave completed a $2.25 billion convertible bond issuance. This financing method has a lower interest rate than the asset-backed financing the company typically uses to pay for new data center construction, but it also carries the risk of diluting shareholder equity and depressing stock prices. The cost of default insurance for CoreWeave's debt has soared to 7.9 percentage points.

Trust Crisis in AI Infrastructure Investment

CoreWeave's turmoil highlights broader issues affecting the entire AI industry: rapid growth has raised questions about when and how significant capital investments will generate healthy profits.

Notable short-seller Jim Chanos, known for accurately predicting the Enron collapse, has publicly criticized CoreWeave, intensifying market concerns about an AI bubble. At the end of October, after hedge fund Two Seas Capital publicly opposed it, Core Scientific shareholders overwhelmingly rejected CoreWeave's proposed $9 billion acquisition offer. Two Seas wrote in a letter that if the deal went through, Core Scientific shareholders would be "exposed to the high volatility of CoreWeave's stock price," facing "significant economic risks." After the deal failed, CoreWeave's stock price dropped over 6%.

**Concerns about construction delays have spread throughout the industry. In recent earnings reports, cloud service provider Oracle and custom chip designer Broadcom both faced double-digit percentage declines in stock prices after indicating that some expenditures would be later than investors expected. Broadcom's stock fell a cumulative 18% over three trading days, marking its worst performance over that period since March 2020, with a market value evaporating by over $300 billion. After Oracle reported unexpectedly high capital expenditures, the bond market experienced turbulence, raising the capital costs for many large tech companies **

Sina Toussi, founder and Chief Investment Officer of Two Seas, stated that the current market concern is "the long-term value of AI." He also holds shares in CoreWeave and mentioned that the company "excels at quickly launching large workloads with maximum utilization and swiftly replacing underperforming nodes without interrupting workflows," but acknowledged that current market sentiment is cautious.

Construction delays and supply chain bottlenecks across the industry threaten to postpone spending plans worth hundreds of billions of dollars, which have already been factored into valuations