Nasdaq 100 Slips as $110 Oil and a 14.5% Bond Yield Expose the AI Boom's Funding Gap

A barrel of Brent crude above $110 and a data-centre operator paying 14.5% to borrow are not obviously connected. For the Nasdaq 100, they are two sides of the same problem. The index fell 2.5% to 29,203.29 points as investors rotated out of technology and into defensive healthcare names, which notched record highs. The selling was concentrated in the market's former growth engines — software and semiconductors among them — so this was no indiscriminate exit from equities. It was a change of yardstick.

Oil gave the reassessment its trigger. Brent crossed $110 a barrel for the first time since May, according to NBC News, on fears of supply disruptions in the Middle East. Higher energy costs threaten to prolong price pressure just as investors were hoping for cheaper financing. JPMorgan data put US consumer prices up 3.7% year on year in September, with gasoline 9.1% more expensive than in the prior month and responsible for more than half the monthly increase.

Rate expectations offered no clean signal. Media reports suggested traders were increasingly betting on unchanged Fed rates in October — a pause, not a cut. That distinction matters for capital-intensive technology companies, whose spending plans lean heavily on credit.

The credit market speaks first

CoreWeave shows how quickly that dependence can turn into a liability. The Wall Street Journal reported concerns that AI infrastructure firms may not service their debt from operating earnings in time. CNBC put the yield on one CoreWeave bond at about 14.5%, against 9.75% when it was issued in April. Reuters reported yields above 15%. Either way, the message from the credit market is the same: risk is being repriced, and the pressure has spread from equities to the financing of the data centres themselves.

That reframes the question hanging over the index. It is no longer whether demand for artificial intelligence exists, but whether the returns justify the outlay. Goldman Sachs analysts, cited by Reuters, expect five large AI hyperscalers to lift combined capital spending to roughly $1.2 trillion next year — a colossal build-out that says nothing yet about the profits it will generate. Broadcom CEO Hock Tan, speaking in Singapore, took aim at inflated expectations among AI investors, Fortune reported, turning attention to the economics of the boom rather than its pace.

Record results, falling shares

Earnings season has already shown how high the bar has risen. ASML posted record quarterly revenue and TSMC a record profit; neither stock held its gains. Good numbers alone are no longer enough to dispel doubts about the next stage of AI investment.

The coming weeks bring more tests. Meta and Tesla are due to report on 21 October, with Lam Research and Texas Instruments also in focus. Bank of America expects Meta revenue of $63.8 billion against a consensus of $63.3 billion, leaving room for an operating surprise — but the bank also raised its capital expenditure estimate for the company. Its 2027 investment forecast now stands at $205 billion, up from $192 billion, an analyst projection rather than company guidance.

Growth and rising funding needs will therefore sit side by side in the same reports. What investors want is the link between the two: evidence that spending discipline and the returns on added capacity can carry the AI cycle. Until oil and financing costs stop pushing in the same direction, strong revenue figures on their own are unlikely to settle the argument.