Is Nvidia too big to fail? AI-driven funding web raises risks

Suspicions of sex and spying swirl Nvidia CEO Jensen Huang is raising an alarming question: Could trouble at the chipmaker spread through the tightly connected financial network that NVIDIA has built to support the artificial intelligence boom?

A significant problem with Nvidia could result in widespread economic and financial harm. The company has provided hundreds of billions of dollars in funding to other companies to help grow the industry, and has been dubbed the “central bank of AI.” The Federal Reserve and other financial institutions are closely monitoring the build-out of AI to monitor the potential for unforeseen events to destabilize the global financial system.

As recently as July, Fed officials were discussing the possibility of “broad asset repricing” if problems arose in the AI ​​sector.

Analysts and current and former financial officials say a repeat of the 2007-2008 financial crisis is unlikely. Sheila Baer, ​​a former government regulator who ran the Federal Deposit Insurance Corporation, said Nvidia is not “too big to fail in the banking sense.” Unlike banks that fail, NVIDIA is not immune to sudden loss of credibility and will continue to produce products that customers want to buy even if something goes wrong. And that huge amount of capital provides a cushion.

But regulators need to closely monitor the potential exposure of banks and the broader financial system to Nvidia and similar companies, Baer said.

“If NVIDIA went bankrupt, it would be quite a shock event,” she says.

Investors still have high hopes for Nvidia, which will report earnings on Tuesday. Analysts expect the company’s sales to more than double this fiscal year and roughly double again over the next two years, according to LSEG data. But with Mr. Huang under intense scrutiny from the FBI, Congress and Nvidia’s board over his relationship with a woman who the New York Times reported was a suspected Chinese spy, analysts say a variety of outcomes are suddenly possible, including the possibility that Mr. Huang, who co-founded the company in 1993, could be fired.

The company did not respond to requests for comment. Nvidia’s board has expressed confidence in its CEO.

Nvidia’s stock price fell 30% from its October 6 high to Tuesday’s close, amid a widespread reassessment of AI trade following Democrats’ surprise victory in the Nov. 3 midterm elections. The company’s stock rebounded Wednesday, rising more than 4% as of midday. Still, people who closely track the financial flows underlying AI say even larger losses for Nvidia are unlikely to cause systemic problems that would alarm regulators.

Still, the risks posed by AI are keeping some finance officials up at night. They are watching closely for the possibility that many of the factors that have fueled the AI ​​boom will simultaneously reverse. For example, if the value of Nvidia’s GPU chips falls, investors may begin to question whether the companies that borrowed money to buy the GPU chips will be able to repay their debts.

In that case, the financial support big tech companies have been providing to their peers could suddenly become a drag on growth. High-tech companies will face credit problems at the same time as their capital falls. Problems among big tech companies can cause problems for the entire economy, not just for tech investors’ portfolios.

“I think this is really concerning because the economy and the financial markets are both on the same footing, which means the drawdown could have a reinforcing effect between them,” said Jason Wu, assistant director of global market analysis at the International Monetary Fund.

Wu was the lead author of the IMF’s Global Financial Stability Report, published in October, which warned of the potential for growing influence among AI companies and investors to cause broader problems. So-called hyperscalers, the largest tech companies, have grown from just 1% of bond issuance in the $15 trillion investment-grade bond market to 10% in just two years. Hedge funds and individual investors are making even riskier bets by borrowing money to buy stocks and investing in leveraged funds.

Nvidia is at the center of it all. The company has repeatedly stepped in to help backstop a small AI company known as Neocloud, which uses Nvidia chips to build data centers and leases capacity to other companies. Among other obligations, NVIDIA disclosed in July that it had committed to purchasing up to $36 billion in cloud services from these companies. This will ensure Nvidia has more customers for its AI chips and make it easier for those companies to raise capital.

Wu and his colleagues refer to this type of arrangement as circular financing. When a business is growing, it is productive for large companies to help customers purchase products. But if Nvidia fails to meet its obligations, there will be repercussions for the companies that borrowed the money and the banks that made the loans. Relatively opaque private credit funds are also lending heavily to the AI ​​boom, and their fate is the subject of debate among regulators.

Nvidia did not reveal specific partners, but some neoclouds, such as CoreWeave and IREN, have separately disclosed relationships with larger companies.

Many neo-crowds have been under stress in recent weeks. The cost of insuring CoreWeave debt, as measured by credit default spreads, rose 94% from Oct. 1 to Nov. 5 before settlement, nearly doubling, according to LSEG data. The price of IREN convertible bonds due 2029 fell 23% from Nov. 3 to Nov. 5, while the company’s stock price fell 26%.

Nvidia’s other obligations include a $105 billion tiered guarantee, subject to certain conditions, for a major data center project in Ohio planned by OpenAI and SB Energy. Nvidia has pledged to backstop up to $125 billion of the Wall Street financial firm’s $500 billion credit program.

These obligations come in many forms and are unlikely to be due all at once, if at all. Stein van Neuerburg, a finance professor at Columbia Business School, estimates they total $400 billion.

That’s a huge amount, but even after the stock price drop, it’s only a fraction of Nvidia’s $4 trillion market cap. “The company would have to be brought to the brink of extinction before all these guarantees were jeopardized.”

“I’m not saying this is a safe system, but I’m saying there’s a lot of equity in the system right now,” Van Neuerburgh said.

But in recent days, Nvidia’s problems have separated it from its competitors. Most NeoCloud stocks have rebounded this week while Nvidia’s stock price has fallen. This suggests that investors currently view Nvidia’s problems as unique to the company.

Meanwhile, basic laws of economics mean other companies are trying to tap into Nvidia’s highly profitable chip business. AMD, broadcom, intel, google, microsoftand OpenAI are both working to develop their own AI chips, Van Nieuwerburgh said.

Over time, they may replace Nvidia products. That would limit the financial impact of potential problems for the company.

Still, it’s impossible to completely eliminate the financial risk of AI trade suddenly hitting a wall, added Van Neuerburg, speaking to me via Zoom from the sidelines of the bimonthly meeting of financial officials in Basel, Switzerland.

“Central bankers are clearly very, very concerned about this and are thinking seriously about these issues,” he said.