US AI Stocks Lost Trillions in a Week as the Market Starts Pricing in 'Borrowing to Buy Shovels'

Original Author: Ada, Deep Tide TechFlow

This past week, the US stock AI sector presented a strange picture: records were broken one after another, while stocks were sold off one after another.

On June 1, Alphabet, sitting on a hundred billion dollars in cash, announced one of the largest equity financings in history. On June 3 after market close, Broadcom delivered its best-ever quarterly report, only to see its stock price crash the next day. On June 5, the Nasdaq composite index fell 4% in a single day, and the semiconductor sector evaporated about $1 trillion in market value. On June 10 after market close, Oracle reported record revenue and backlog, yet its stock price still fell. On June 11, the史上 largest SpaceX IPO entered its pricing moment. The earnings numbers themselves are not the problem; the problem is how these numbers were achieved: more and more money is being borrowed in increasingly complex ways to fund this AI infrastructure race. When the market starts to count the cost, even record numbers can't save the stock price.

The Same Script: First Set a Record, Then Get Hammered

Broadcom's script came first. According to the company's earnings and multiple media reports, for the fiscal second quarter ended May 3, Broadcom reported revenue of $22.219 billion, up 48% year-over-year, AI chip revenue of $10.8 billion, up 143% year-over-year, and earnings per share that beat Wall Street expectations. However, the market focused on one gap: the company's guidance for next quarter's AI chip revenue was $16 billion, below analyst expectations. CEO Hock Tan also did not raise the full-year AI revenue target and mentioned that Google might diversify its chip supply chain. The next day, Broadcom's stock price fell about 15%, evaporating nearly $280 billion in market value in a single day, one of the largest single-day market value evaporations in Wall Street history.

A week later, it was Oracle's turn. According to the company's earnings and CNBC, for the fiscal fourth quarter ended May 31, Oracle reported revenue of $19.2 billion, up 21% year-over-year, cloud infrastructure revenue of $5.8 billion, up 93% year-over-year, and adjusted earnings per share of $2.11, above the analyst estimate of $1.95. The backlog was even more staggering: remaining performance obligations (RPO) reached $638 billion, a massive year-over-year increase of 363%, far exceeding the analyst estimate of $595.7 billion. But the stock price again refused to buy in, falling about 9% in after-hours trading.

Sandwiched between these two earnings reports was the widespread sell-off on June 5. According to TheStreet and CNBC, the Nasdaq Composite Index fell 4% that day, its worst single-day performance since the tariff turmoil in April 2025. Triggered by Broadcom's cautious outlook on AI chips, AMD and Intel led the decline in the entire semiconductor sector.

It's worth noting that the crash on June 5 wasn't entirely about "AI skepticism." On that day, the US nonfarm payrolls increased by 172,000, far exceeding expectations, pushing up interest rate expectations and causing funds to rotate from high-valuation growth stocks to defensive sectors like healthcare and consumer staples. AI stocks, having the highest valuations, fell the most. In other words, macro interest rates and sector rotation were one driving force, while concerns about AI capital expenditure were another. The two overlapped, rather than being attributable to a single cause.

What Got Hammered Wasn't the Income Statement, But the Cash Flow Statement

Piecing together the three market movements reveals a common point: while the income statement was still recording "records," the market had already shifted its focus to the cash flow statement and balance sheet. The center of pricing gravity moved from "how much was earned" to "how much more needs to be burned and borrowed to earn this."

Oracle is the most straightforward example. According to the company's earnings report, operating cash flow for the full fiscal year 2026 reached a record $32 billion, up 54% year-over-year, but free cash flow was negative $23.7 billion. The company raised $43 billion in debt and $5 billion in equity during the fiscal year. What truly broke market sentiment, however, was its forward-looking narrative. According to CNBC, Oracle plans to raise approximately $40 billion more in fiscal year 2027 through a combination of debt and equity. A company that just raised nearly $50 billion and has negative free cash flow, now pre-announcing another $40 billion round, when paired with the word "record," led the market to price the latter.

Broadcom's logic is similar, though manifested differently. According to Barron's, Broadcom lowered its gross margin guidance for the third fiscal quarter from 77% to 74%, attributing it to the rising share of lower-margin AI chips in its revenue mix. Coupled with a retreat from selling "full systems" to just selling "chips," and customers demanding to lease chips (passing on financing pressure), the market sees a business growing rapidly but with deteriorating margins and capital intensity.

Goldman Sachs provided a framework for this shift. According to its research report, investor tolerance for rising capital expenditure depends on the strength of earnings and the visibility of AI monetization. The same report noted that Alphabet's stock price rose after it raised its profit guidance, while Meta's fell after its guidance remained flat. The market no longer rewards "growth" indiscriminately but distinguishes winners from losers based on the ability to "monetize."

The Financing Chain Takes Center Stage: Even the Player with the Most Cash is Borrowing

If the income statement is the surface, the financing chain is the true protagonist of this week. From the very upstream to the very downstream, almost every link is paying for the same AI infrastructure buildout by leveraging debt or diluting equity.

The most compelling example is Alphabet. According to its SEC filing, on June 1, it announced an $80 billion equity financing, which was upsized and priced at $84.75 billion on June 2, including a $10 billion private placement from Berkshire Hathaway. The anomaly is that this company is not short of cash. According to multiple media reports, as of March 2026, Alphabet held $126.8 billion in cash, with $174 billion in annual operating cash flow, and had already issued over $55 billion in new debt since November. Even so, Melius Research estimates that Google's free cash flow will turn negative in the coming years. Commenting on this, investor Dan Niles noted that capital is not infinite, and the fact that Google, with "the strongest AI tech stack out there," still needs to raise massive funds precisely illustrates the intensity of this spending cycle.

Moving downstream, every link in the chain is doing the same. New cloud player Oracle has negative free cash flow, relying on debt and equity financing, and asks customers to prepay for GPU purchases or bring their own GPUs to reduce its own upfront construction costs. Chip supplier Broadcom, on June 9, partnered with Apollo and Blackstone to launch the AI XPV Platform, initially capitalized at $35 billion, aiming to support over 20 gigawatts of computing power by 2028, serving frontier labs including Anthropic and OpenAI. At the very end of the chain, the labs themselves are deploying even more aggressive tools: previous reports indicated SoftBank arranging margin loans backed by OpenAI equity. Now, SpaceX is racing towards a Nasdaq IPO targeting a $75 billion valuation, Anthropic has filed confidentially for an IPO, and OpenAI is expected to follow.

The total amount of this spending is also rapidly expanding. According to CreditSights, total capital expenditure for hyper-scalers in 2026 is estimated at around $750 billion, an increase of approximately 67% from 2025. Another metric from Goldman Sachs shows that their 2026 CapEx forecast has been revised up from $314 billion at the start of the year to $518 billion. Regardless of which metric is used, the direction is the same: spending is accelerating, while the portion that can be covered by operating cash flow is shrinking, and the gap must be filled by the capital markets.

The Fulcrum of the Chain Rests on a Few Labs That Aren't Profitable Yet

Leverage itself isn't scary. What's scary is who the leverage ultimately depends on. Tracing this financing chain to its end reveals a highly concentrated fulcrum.

Oracle's $638 billion backlog might seem unshakeable, but according to Bank of America, over 50% of it comes from a single customer: OpenAI. Oracle has also disclosed that most of the RPO growth in the past two quarters came from large AI contracts where customers either prepaid for GPUs or procured GPUs themselves to entrust to Oracle. Broadcom's six key custom chip customers are similarly concentrated among a few players like Google, Meta, Anthropic, and OpenAI. In other words, from the financing of hyper-scalers, to the orders of tool-sellers, to the injection of private credit and insurance capital, the ultimate payer for the entire chain converges on a handful of frontier labs like OpenAI and Anthropic—labs that are not yet profitable and are themselves queuing up for financing.

The record revenue is real. The $638 billion backlog is also real. But the payer for these orders is highly concentrated, itself relying on financing to survive. The leverage on the entire chain is being reassessed by the market. This week, the market didn't deny AI's growth; it just started demanding to see who pays for this growth bill, and how. SpaceX is set to price after the market close on June 11 and list on Nasdaq on June 12 at $135 per share, targeting a valuation of approximately $1.77 trillion. Whether this史上 largest IPO can be smoothly digested will be the next stress test for this financing chain.