×

Analysis-Nvidia turmoil marks the end of easy money for the AI trade

By Thomson Reuters Nov 17, 2026 | 5:12 AM

By Gertrude Chavez-Dreyfuss

NEW YORK, Nov 17 (Reuters) – Tumult at Nvidia spells the end of easy money for the AI sector and a funding squeeze for smaller, highly leveraged “neocloud” companies that have helped fuel the industry’s rapid expansion.

In recent years, Nvidia’s financial support has boosted infrastructure providers and other partners of the giant chip firm. Bond investors say the uncertainty over who might eventually succeed ousted Chief Executive Jensen Huang ​has raised questions about Nvidia’s willingness to continue support through vendor financing, guarantees and other methods, hitting these firms’ shares and bonds.

“It’s fair ‌to say that anyone who takes over is going to have a lower risk threshold than Jensen when it comes to Nvidia’s balance sheet,” said Loren Moran, fixed income portfolio manager, at Wellington Management. “There may just be a lot less credit availability toward weaker players.”

Chief among the firms cast in an uncertain new light are the so-called neocloud firms such as CoreWeave and Nebius, cloud computing companies that acquire graphics processing units (GPUs) or electronic chips, renting out computing capacity to AI developers, startups, and large technology companies.

NEOCLOUD BONDS DROP

Last week, credit investors resumed their retreat from bonds ‌exposed ​to Nvidia-related risk. Bonds in CoreWeave issued in June at a yield of 9.625% plunged in value, sending ⁠their yield to nearly 15%. The company’s shares ⁠have lost 45% since the Nasdaq hit its latest record on October 6, even after the company said on its earnings conference call last week that it would be able to continue to finance its operations by doing things such as borrowing against its chips.

Nvidia is a shareholder and buyer of last resort for CoreWeave’s unsold cloud capacity through April 2032 under a $6.3 billion order.

Convertible bonds of Nebius, issued at a 4.5% yield in August, dropped ​to 70 US cents to the dollar last Friday, sending yields surging to more than 12%. Nebius rents out the computing power that AI firms need to train and run their models. Nvidia said in March it would invest $2 billion in Nebius, representing an 8.3% stake in the company. Nebius stock is down 57% ⁠since October 6.

Many of these neocloud companies operate with smaller balance sheets while relying heavily on ⁠external financing to fund the enormous costs associated with acquiring GPUs and building data centers.

“Nvidia’s contracts and existing guarantees should ​not change with Jensen Huang’s departure,” said Tony Trzcinka, investment-grade portfolio manager at Impax Asset Management. “It’s all about whether Jensen’s successor would want to be as adventurous in ​going into all these different arrangements.”

WIDENING DISPARITY

For Nvidia itself, some investors argued that a more conservative approach to balance sheet risk ‌could be credit-positive. Wellington’s Moran said she was concerned about the company’s huge financial commitments across the AI space and believes a reduced appetite for risk could “clip some of that tail.”

But a more cautious Nvidia could have different implications for the broader AI financing environment. It could, for instance, widen the disparity between investment-grade borrowers and lower-rated companies dependent on private credit, leveraged loans and high-yield debt markets. Companies relying on financing tied directly or indirectly to Nvidia may find future capital either harder to ⁠obtain or significantly more expensive, analysts said.

Hyperscalers such as Alphabet and Microsoft appear best positioned to weather such a shift, investors say, given their strong balance sheets that can finance data center construction on their own.

“Most of their AI spending comes from internal cash flow and debt markets,” said Danny Zaid, portfolio manager at ⁠TwentyFour Asset Management, and that is “unlikely going to change in ‌the near term with the strong AI momentum expected to continue.”

At the other end of the spectrum, investors have ⁠already been demanding greater compensation to finance data center projects from lower-rated AI borrowers. “In a construction phase (of a project) ​where there is ‌a high-yield name, you’re going to have a materially higher cost of funding,” said Trzcinka of Impax.

In Goldman ​Sachs’ “Top of Mind” research ⁠note, Miriam Wheeler, the bank’s global head of leveraged finance, said the pace of AI deals in the high-yield market has significantly slowed in recent weeks amid investor concerns over mounting supply and political uncertainty.

Wheeler said the market had expected roughly $60 billion of issuance this fall, which had pressured spreads. But supply has dried up since late September, “with no new deals over the past month.”

The slowdown suggests the easy financing conditions that have fueled AI-linked issuance are beginning to fade, analysts said.

“The uncertainty for the overall space means greater risk premiums expected by investors,” said Zaid of TwentyFour Asset Management. “Expect dispersion to increase as investors evaluate credits on a case-by-case basis.”

(Reporting by Gertrude Chavez-Dreyfuss, ​editing by Colin Barr and Nick Zieminski)