AI Giants’ Debt-Fueled Buildout Tests Market Power

CoreWeave, an AI cloud company with about $50 billion in debt, saw its shares tumble as investors questioned who can keep financing the AI build-out—and whether companies borrowing for data centers can make them pay. The market’s uncertainty has reached the heart of a bull market marking its fourth anniversary, putting corporate earnings and debt under scrutiny.

Who Pays for the Buildout

Investors are weighing whether companies pouring money into AI infrastructure can generate returns sufficient to sustain their spending. That question puts corporate borrowing plans, along with the lenders and investors backing them, under pressure. The base article describes market actors and corporate finances; it identifies no community-led response or mutual-aid effort.

“The narrative in the market is a little bit jittery,” said Garrett Melson, portfolio strategist with Natixis Investment Managers Solutions. “Clearly there’s a lot of uncertainty around where AI stands right now.” Semiconductor companies and technology hardware firms, including Dell Technologies, recorded sharp declines. The S&P 500 remained about 1% below its record high set that month, while sectors that had lagged technology shares began picking up the slack.

“When AI is working, it sucks all the oxygen out of the room for the rest of the sectors,” Melson said, describing an on-off market in which AI’s fortunes can dominate attention and money. The Philadelphia SE Semiconductor index had fallen 8% that month but still stood up more than 60% for the year, compared with a 13% year-to-date gain for the S&P 500.

Debt, Ratings and the Cost of Capital

Deutsche Bank and RBC Capital Markets separately cut their recommendations on the technology sector. On October 9, Deutsche Bank strategists said the rotation into technology had “run far and the risk-reward no longer looks favorable in the near term.” Appetite for AI shares now sits alongside doubts about the debt and spending needed to build the infrastructure.

A Financial Times News Briefing transcript reported that Anthropic received a double-B grade from Kroll Bond Rating Agency in June, based on documents FT reporters saw. The rating, not previously reported publicly, put the AI lab close to investment grade. KBRA lacked full access to Anthropic’s financial information, and the agency used an in-house credit risk score after receiving only a limited set of financials. KBRA and Anthropic declined to comment.

The transcript said bankers acting for Anthropic had lobbied for an investment-grade rating after the company goes public. Michelle Chan, the FT’s US credit correspondent, said such a rating could lower borrowing costs, help lenders assess creditworthiness and open access to a larger pool of capital. She said Anthropic appeared close to investment grade under KBRA’s framework if it had a substantial cash cushion after an IPO, while noting KBRA isn’t one of the three major credit rating agencies and others might give the company less leeway.

The Return Question

Goldman Sachs estimated Meta and four other AI hyperscalers together would raise capital spending to about $1.2 trillion next year. Investors want evidence those companies can earn returns on that spending. “One of the most important takeaways of earnings season is whether or not the profitability of the hyperscalers can grow faster than the pace of their capex,” said Kevin Shea, senior equity analyst at BNY Wealth.

Meta, with a market capitalization of about $1.9 trillion, saw its shares rise in recent weeks after launching its Muse AI assistant. “With the cost of capital going up, this capex blowout is going to become more expensive, which raises the bar for returns,” said Chuck Carlson, chief executive officer at Horizon Investment Services. Meta was due to report among companies facing questions about the payoff from the buildout.

The Financial Times’ Katie Martin said South Korea’s semiconductor-heavy stock market fell 8% in a session, with selling carrying into Europe and, to some extent, the United States. She said CoreWeave wanted an investment-grade rating for its bonds, which a senior executive said could help the company tap insurance-company capital. Martin described the AI trade over the previous 12 to 18 months as driven in large part by “hype and vibes,” and said it remained unclear whether the pullback was a short-term wobble or the start of a more serious downturn.

Reporting season was expected to heat up, with S&P 500 companies forecast to increase third-quarter earnings by more than 30%, according to LSEG IBES. Meta, Lam Research and Texas Instruments were among companies preparing to report, alongside Tesla, Netflix, Procter & Gamble and GE Aerospace. Rising Treasury yields and possible further Federal Reserve rate increases remained concerns for stock investors. The earnings question is straightforward: can companies’ returns catch up with the vast sums committed to building their AI business?