A massive upheaval hits the "AI central bank", how far will the impact ripple along the financing chain?

The sudden resignation of Jensen Huang, CEO of NVIDIA, has raised a worrying question: if the chip giant runs into trouble, will the shock spread through the interconnected financial network it has built to underpin the artificial intelligence (AI) boom?

To drive the development of the AI industry, NVIDIA has made hundreds of billions of dollars in financial commitments to other enterprises, earning it the title of "central bank of AI". As a result, once NVIDIA encounters severe problems, it may have a far-reaching impact on the U.S. economy and financial markets.

Back in July, officials at the Federal Reserve, the real central bank of the United States, had already discussed whether there would be a "widespread revaluation" of asset prices if problems arise in the AI industry.

Regulators Focus on Risk Exposure

Analysts, current and former financial officials believe that a repeat of the 2007-2008 financial crisis is unlikely.

Sheila Bair, a former regulator who once headed the U.S. Federal Deposit Insurance Corporation (FDIC), believes that from the perspective of banking regulation, NVIDIA is far from being "too big to fail".

NVIDIA is not as vulnerable to sudden loss of confidence as troubled banks. Even if something goes wrong, it can still produce products that customers are willing to buy, and its huge equity value also provides a buffer for it.

However, Bair pointed out that regulators need to pay close attention to the exposure of banks and the entire financial system to NVIDIA and similar enterprises. "If NVIDIA falls, it will be a pretty big earthquake."

Investors still have high hopes for NVIDIA, and the company will release its financial report next Tuesday. According to LSEG data, analysts expect NVIDIA's revenue in this fiscal year to more than double that of the previous fiscal year, and nearly double again in the following two years.

But Jensen Huang's departure has made the company's prospects exceptionally unpredictable. Since co-founding NVIDIA in 1993, Jensen Huang has led the company. As of Thursday morning, NVIDIA's stock price has fallen by more than 30% from its peak on October 6.

However, people who closely track the capital flow of the AI industry believe that even if NVIDIA's stock price plummets further, it is unlikely to trigger the kind of systemic problems that regulators are guarding against.

If GPUs Depreciate, Who Will Repay the Debts for Purchasing Chips?

Nevertheless, the risks brought by the AI boom still keep some financial officials awake at night.

They are concerned about whether multiple factors driving this boom will reverse at the same time. For example, if the value of NVIDIA's GPU chips falls, investors may begin to doubt whether enterprises that borrowed money to buy chips can repay their debts.

At that time, the financial support provided by large technology companies to other enterprises in the industry may suddenly become a drag on their own growth. Technology companies will face both credit pressure and falling stock prices. Once tech giants get into trouble, the shock may spread from tech stock portfolios to the entire economy.

"I think this is really worrying, because both the real economy and the financial market rely on the same growth pillar. Once the market falls, the two may drag each other down and amplify the impact," said Jason Wu, assistant director of the Global Markets Analysis Division at the International Monetary Fund (IMF).

Wu is the lead author of a chapter in the IMF's Global Financial Stability Report released in October. The chapter warns that the continuous leverage increase by AI enterprises and investors may trigger broader problems.

In the $15 trillion investment-grade bond market, "hyperscale cloud service providers" (the largest few technology companies) have seen their share of bond issuance rise from 1% to 10% in just two years.

Hedge funds and retail investors are also further amplifying already risky bets by borrowing money to buy stocks or investing in funds with embedded leverage.

From Selling Chips to Purchasing Computing Power: How NVIDIA Supports "Circular Financing"

NVIDIA is at the center of this financial network. It has repeatedly supported smaller AI cloud service providers. These enterprises, known as "neoclouds", use NVIDIA chips to build data centers and then rent computing power to other companies.

NVIDIA disclosed in July that its commitments include purchasing up to $36 billion of cloud services from such enterprises. This not only helps NVIDIA expand its customer base for AI chips, but also makes it easier for these cloud service providers to obtain financing.

Wu from the IMF and other researchers call such arrangements "circular financing". When business is growing, large companies helping customers buy their own products can drive business development. But if NVIDIA fails to fulfill its commitments, the shock will be transmitted to borrowing enterprises and related lending banks.

Private credit funds with relatively low transparency have also provided a large number of loans for this AI boom. How much impact these funds will suffer once the market reverses is still a topic for discussion among regulators.

NVIDIA has not disclosed specific partners, but neocloud service providers such as CoreWeave and IREN have each disclosed their partnerships with NVIDIA.

In recent weeks, many neocloud service providers have faced pressure. According to LSEG data, from October 1 to November 5, the cost of debt default protection for CoreWeave, measured by credit default swap (CDS) spreads, rose by 94%, nearly doubling, before stabilizing.

From November 3 to 5, the price of a 2029 maturing convertible bond of IREN fell by 23%, while the company's stock price fell by 26% in the same period.

With around $400 billion in commitments, how big is NVIDIA's buffer?

NVIDIA's other commitments include providing a total of $105 billion in guarantees in phases for a large data center project planned by OpenAI and SB Energy in Ohio, under specific conditions.

NVIDIA has also committed to providing up to $125 billion in backstop support for a $500 billion credit program for Wall Street financial institutions.

These commitments take different forms, some of which may not need to be actually fulfilled, and even if triggered, it is unlikely that they will mature at the same time. Stijn Van Nieuwerburgh, a professor of finance at Columbia Business School, estimates that these commitments add up to about $400 billion.

Nieuwerburgh said: "The company would have to be on the verge of life and death for all these guarantees to be threatened. I would not conclude that the system is safe. But right now, there is indeed a lot of equity capital in the whole system."

The Chip Market Landscape May Face Reshuffling

In the past few days, NVIDIA's stock price trend has diverged from that of most neocloud service providers. NVIDIA's stock price has fallen this week, while the stock prices of most neocloud service providers have rebounded. This shows that, at least for now, investors believe the problem lies mainly with NVIDIA itself.

At the same time, the huge profits in the chip business are attracting other enterprises to enter the market where NVIDIA operates. Nieuwerburgh said that AMD, Broadcom, Intel, Google, Microsoft and OpenAI are all developing self-developed AI chips.

Over time, these chips have the potential to replace NVIDIA's products, thereby limiting the economic impact if NVIDIA runs into trouble.

But Nieuwerburgh added that the financial risks brought by a sudden collapse of the AI investment boom cannot be completely ruled out. He said that central bank officials in various countries are "obviously very, very, very anxious and are seriously thinking about these issues".

This article is from the WeChat official account "CLS.cn", written by Zhao Hao, and published with authorization by 36Kr.