AI Stock Sell-Off: $80 Billion Wiped Out as US Inflation Fuels Rate-Hike Fears

The AI boom has a financing problem to solve alongside its technology challenge. CoreWeave, Nebius, Oracle and Dell sold off sharply on October 14 as hotter US inflation added to concerns about borrowing costs. Reuters estimated that the four companies had lost roughly $80 billion in market value during trading. Our assessment: investors are becoming more demanding about how AI expansion is funded and how much cash it will eventually produce.

Let’s break down what triggered the AI stock sell-off, why these four businesses face different risks and which financial numbers investors should watch next.

How much did CoreWeave, Nebius, Oracle and Dell fall?

The selling hit both companies that rent AI computing capacity and companies that supply the hardware. CoreWeave recorded the steepest percentage decline among the four.

These closing moves differ from the rounded percentages in the intraday Reuters report behind the headline. Keeping those timestamps separate avoids presenting an afternoon market-value estimate as an exact end-of-day calculation.

Market value is the share price multiplied by shares outstanding. The decline reduced the stock market’s valuation of these businesses. It did not mean that the companies collectively spent or lost that amount of cash in one day.

However, a lower share price can affect future financing. A company raising a fixed amount of money by issuing shares must issue more shares when its price falls, increasing dilution for existing shareholders.

Why did US inflation trigger an AI stock sell-off?

The September inflation report challenged expectations that price pressures would ease sufficiently to limit further Federal Reserve tightening.

Source: US Bureau of Labor Statistics, September CPI release published October 14. 

Gasoline accounted for more than half the monthly headline increase. Core inflation also moved higher on an annual basis, which made the report harder to dismiss as solely an energy-price shock.

The Fed had already raised its policy rate in September. Its next scheduled meeting is October 27–28. The latest CPI release intensified concern that another increase could follow, but a hotter inflation report does not guarantee a particular decision.

The Fed’s formal longer-term inflation objective uses the personal consumption expenditures price index (PCE). CPI is a closely watched input into the policy outlook rather than the measure that defines that objective.

Sources: Federal Reserve’s September 16 policy statement, meeting calendar and inflation FAQs.

Why AI infrastructure is especially sensitive to interest rates

AI infrastructure requires substantial spending before all the associated revenue arrives. Providers purchase chips, secure power and build data centres in anticipation of customer usage over future periods.

Higher financing costs can affect these projects in three connected ways:

1. New borrowing becomes more expensive. More of a project’s eventual earnings must go to lenders.
2. Future cash flows become less valuable today. Investors demand a higher return before committing money to a long-term project.
3. Customers face tougher spending decisions. A higher financing hurdle can delay projects even when interest in AI remains strong.

The exposure depends on who funds the investment and when the cash arrives. Companies with substantial existing cash generation have more flexibility than businesses that must repeatedly seek outside capital.

There is also a difference between inflation-driven market pressure and a company’s own credit risk. Investors can demand a higher return for lending to a particular borrower even when Treasury yields are falling.

Did Treasury yields finish higher after the inflation report?

Long-term yields initially rose before retreating. The final daily Treasury readings show that the movement was different across maturities.

Shorter-term yields rose while longer-term yields finished below the previous day’s reference levels. Long-term financing nevertheless remained expensive and concerns about individual AI borrowers persisted.

The sell-off therefore reflected a combination of inflation uncertainty, demanding valuations and company-specific funding concerns. Describing it simply as a day when every bond yield rose would miss part of the story.

Why these four AI stocks face different financial risks

The four companies participate in different parts of the AI supply chain. Their latest reported results help explain why a shared market shock can create different business consequences.

Sources: The companies’ latest quarterly earnings releases and Nebius’s shareholder letter. CoreWeave and Nebius quarters ended June 30, Oracle’s quarter ended August 31 and Dell’s quarter ended July 31. 

The key distinction is between generating revenue and funding the assets needed to earn it. Cloud infrastructure owners must finance capacity. Hardware suppliers must convert customer demand into deliveries, margins and cash collection.

CoreWeave: The bond market is questioning funding risk

CoreWeave rents specialised computing capacity to customers developing and running AI systems. Its growth requires substantial investment in infrastructure.

Its latest quarterly net interest expense was equivalent to approximately 24.9% of revenue, calculated from the figures above. That is a significant accounting expense before shareholders receive any residual profit.

The company also reported a 59% adjusted EBITDA margin. EBITDA excludes interest, tax, depreciation and amortisation, with further company-defined adjustments. The gap between that margin and the reported net loss shows why an operating metric that excludes financing and asset costs cannot establish the return available to shareholders.

The immediate market concern came from its bonds. Barron’s reported that CoreWeave’s 2032 notes, carrying a 9.625% coupon, were yielding approximately 15% in secondary trading.

A bond’s coupon is its contractual interest rate. Its market yield changes as the trading price changes. A higher market yield does not automatically reset the coupon on existing fixed-rate debt, but it can signal a more difficult environment for new borrowing or refinancing.

CoreWeave points to its contracted revenue as support for its obligations. Its June-end backlog was approximately $104 billion, excluding additional commitments added early in Q3. Backlog still requires delivery and service availability before becoming revenue.

Our view: investors should examine the cash remaining after infrastructure costs and financing obligations. A large contract book supports demand visibility, but its headline size cannot answer that cash-flow question by itself.

Sources: CoreWeave’s Q2 release; Barron’s October 14 bond-market report and company response.

Nebius: Rapid growth comes with a substantial funding requirement

Nebius is also expanding an AI cloud business. Its latest quarterly revenue grew 454% from a year earlier, while capital expenditure was almost ten times quarterly revenue.

That spending ratio reflects a major expansion phase. It does not establish that future spending will remain at the same proportion of sales, but it makes the funding plan central to the investment case.

Nebius ended June with approximately $8.0 billion of cash and cash equivalents, excluding restricted cash. That provides flexibility. Customer prepayments, debt financing and equity issuance also feature in its stated funding approach.

The questions are how quickly new capacity begins earning revenue and how much additional capital is needed before it does. Delayed delivery can extend the period during which an asset incurs costs before producing the expected cash.

Nebius should therefore be assessed on its own financing arrangements and execution. A shared “AI cloud” label does not mean it has the same debt structure or refinancing schedule as CoreWeave.

Sources: Nebius’s Q2 financial statements and shareholder letter.

Oracle: Strong cash generation still trails infrastructure spending

Oracle combines established software operations with an expanding cloud infrastructure business. Its existing customer relationships and operating cash generation provide a broader financial base.

Yet its latest quarter illustrates the scale of the investment burden. Subtracting capital expenditure from operating cash flow produced approximately $5.4 billion of negative free cash flow.

Operating cash flow also included roughly $11.4 billion of customer prepayments with a significant financing component. Those advances help fund capacity, but they carry future delivery obligations and should be considered when judging how repeatable quarterly cash generation may be.

Oracle also reported completing a $20 billion common-stock sale during the quarter. That highlights the use of equity alongside other funding sources.

The investor test is whether the expanding asset base produces attractive returns after its financing and operating costs. Software profitability gives Oracle flexibility, while the speed of its infrastructure expansion creates a separate cash requirement.

Source: Oracle’s Q1 FY2027 earnings release and cash-flow schedules.

Dell: The concern includes expectations and the quality of hardware growth

Dell sells the systems used to build AI infrastructure. It therefore faces a different economic model from a provider renting out computing capacity.

Barclays downgraded Dell to Equal Weight from Overweight on October 14. Reporting on the downgrade highlighted caution about valuations and the durability of enterprise hardware spending.

Dell’s latest results nevertheless showed substantial AI server revenue and backlog. The relevant question is how much of that demand becomes profitable deliveries and cash collected.

Hardware revenue includes the value of expensive components. A larger invoice does not necessarily translate into an equally large increase in profit. Investors should therefore examine margins, customer payment terms and working capital alongside sales growth.

Our assessment: Dell’s business deserves a separate analysis from CoreWeave’s credit stress. Its exposure runs through customer spending, order conversion and the expectations already reflected in its share price.

Sources: Reuters and Barron’s reporting on the Barclays downgrade; Dell’s Q2 FY2027 release.

What do Oracle and Dell’s valuations look like after the fall?

A falling share price improves a valuation ratio only if the earnings used in the calculation remain achievable.

Oracle and Dell provide annual adjusted earnings guidance, allowing a straightforward illustration. These are company forecasts rather than reported annual earnings or rolling 12-month analyst estimates.

Sources: Company earnings releases and Stock Analysis historical prices. 

The ratios show what investors are paying for each dollar of forecast adjusted earnings. They do not establish which stock offers better value: growth, debt, reinvestment requirements and the reliability of earnings all matter.

Adjusted earnings also exclude certain expenses. A business can report attractive adjusted EPS while using substantial cash to build infrastructure or fund working capital.

For CoreWeave and Nebius, the analysis needs particular attention to the path from growing revenue to cash available after infrastructure investment and financing. Nebius’s investment revaluations can also affect reported earnings, making a simple headline earnings comparison less informative.

How higher borrowing costs change the financial calculation

Consider an illustrative infrastructure company seeking new debt or refinancing an existing balance.

The extra financing cost reduces the amount left for shareholders even if the project generates exactly the same operating revenue.

The effect would apply as debt is raised, refinanced or repriced under its contractual terms. Applying a current bond-market yield immediately to every dollar of a company’s existing debt would overstate the near-term impact.

This is why funding terms can move an AI stock before any customer cancels an order.

Does the sell-off mean the AI boom is ending?

The evidence supports greater scrutiny of investment returns. It does not establish that demand for AI computing has disappeared.

Gartner added to the debate on October 14 with a warning about the economics of infrastructure spending.

Source: Gartner’s October 14 research announcement.

An impairment is an accounting reduction in an asset’s carrying value when that value can no longer be supported. It differs from a fall in a company’s stock-market value.

Likewise, ceasing independent operations can include an acquisition or consolidation. It does not mean that every affected provider goes bankrupt or that its infrastructure stops operating.

Gartner’s warning is relevant because growing usage alone cannot guarantee an adequate return on every data centre built. Too much capacity, falling rental prices or expensive financing can reduce returns even in a growing industry.

The counterargument is that stronger demand, better utilisation and more efficient infrastructure could improve the economics. The next financial reports need to demonstrate those improvements through cash generation and funding requirements.

What should investors watch after the AI stock sell-off?

The next useful evidence will come from financing announcements and operating results, alongside inflation and central-bank decisions.

A stabilising share price would be useful market information. Evidence that a company can fund and deliver its plans would be stronger business information.

What this means for Indian investors

Holding several AI-related companies may still create exposure to a common risk. A GPU cloud provider, a server supplier and a cloud software company can all depend partly on continued data-centre investment.

Investors should look through the company names and assess that shared spending exposure. A Nasdaq-100 allocation may also overlap with individual technology holdings, depending on the portfolio.

Currency adds another variable. For an Indian investor, a dollar investment’s rupee return reflects both the stock’s performance and the exchange-rate movement. Rupee depreciation can cushion a dollar share-price decline, while rupee appreciation can deepen it.

For Indian businesses planning AI investments, this sell-off is a reminder to evaluate funding costs, customer commitments and cash returns. The market move alone does not establish that an Indian company’s orders or earnings have deteriorated.

Our view: Funding quality will increasingly separate AI winners

The strongest interpretation of this sell-off is that investors are demanding better evidence of returns on AI infrastructure spending.

CoreWeave’s bonds make financing risk especially visible. Nebius’s expansion puts capacity delivery and capital access in focus. Oracle must balance a broader earnings base against heavy investment, while Dell must convert hardware demand into durable profit and cash.

We would give greater weight to secured funding, realistic delivery schedules and cash remaining after investment than to the size of a backlog alone. A large order book is most valuable when fulfilling it creates an attractive return for shareholders.

The next phase of the AI trade will reward companies that can connect demand to cash generation without imposing an excessive financing burden.