Nvidia (NVDA.US) received a $500 billion funding promise from Wall Street, and concerns about AI “circular financing” resurfaced

The Zhitong Finance App learned that Nvidia (NVDA.US) announced on Monday that it has signed a memorandum of understanding with six financial giants: Apollo Global Management, Blackstone, Brookfield Asset Management, Goldman Sachs, and KKR, to establish an independent computing power financing platform with the goal of mobilizing more than 500 billion US dollars of third-party capital over the long term for AI infrastructure construction. Nvidia founder and CEO Hwang In-hoon said in a joint interview: “This is indeed the first time that a technology chip has become an investable asset class.”

The core logic of the financing platform: turning GPUs into “digital real estate”

Nvidia is trying to compare GPUs and data center infrastructure to commercial real estate, toll roads, or other collaterable infrastructure assets. The logic is that because Nvidia hardware is widely used and can be transferred between customers, lenders can reliably underwrite computing power as an asset with long-term revenue-generating capacity.

The central innovation of this plan is to redefine GPUs and data centers as long-term infrastructure assets that generate revenue rather than hardware that depreciates rapidly. Hwang In-hoon explained this idea in an interview: “Fundamentally speaking, the difference in this industry is that computers are now part of infrastructure, like electricity, like the internet, so you have to think about it as infrastructure.”

Under the agreement, six financial institutions will create a “sizeable exclusive pool” for Nvidia customers to provide financing at “very attractive interest rates.” The funds will target hyperscale cloud service providers, cutting-edge AI labs, and enterprises to build data centers and purchase Nvidia hardware. Apollo President Jim Zelter said, “Modern computing has become a scarce, mission-critical asset class with compelling investment characteristics.” BlackRock Chairman and CEO Larry Fink pointed out that AI infrastructure construction will “create attractive long-term investment opportunities.”

The financing structure is mainly debt-based, and all capital comes from third party capital. Nvidia's role is a “matchmaker” rather than a “funder” — helping customers connect capital while ensuring that these funds ultimately flow to Nvidia's chip purchases. Goldman Sachs CEO David Solomon said that the capital market is sending a signal that “a large amount of money can be used to support this construction.”

The premise of this logic is that chips are no longer viewed as rapidly depreciating hardware, but rather as “productive assets” with long-term value. The addition of six financial giants meant that the story was endorsed by Wall Street — Apollo had assets under management of $1.05 trillion, Blackstone over $1.3 trillion, and Brookfield over $1 trillion. BlackRock CEO Larry Fink said the partnership “combines Nvidia's leadership in accelerated computing with BlackRock's ability to connect long-term capital to critical infrastructure.” Goldman Sachs CEO David Solomon described this moment as an important point in the “historic AI investment cycle.”

From “Stargate” to 500 billion: the evolutionary history of Nvidia's “general contractor”

This $500 billion financing plan is not an isolated incident, but an inevitable result of Nvidia's evolution from a chip supplier to a “general contractor” for AI infrastructure.

As early as January 2025, the Trump administration announced the “Stargate” program — led by OpenAI and SoftBank, to invest 500 billion US dollars to build AI infrastructure over four years. As a core chip supplier, Nvidia immediately began to lay out a broader financing network.

Since this year, Nvidia has been intensively promoting a series of large-scale deals: reaching an AI infrastructure partnership of more than 500 billion US dollars with South Korea's SK Group; discussing financing guarantees of up to 250 billion US dollars for its rental data center, while also discussing financing the purchase of chips worth 350 billion US dollars; investing 2 billion US dollars each in CoreWeave and Nebius; and making “huge” investments in AI startup Safe Superintelligence. Microsoft has both an investment and revenue share in OpenAI. Amazon and Google have invested heavily in Anthropic, while Anthropic has promised to invest large sums of money back into AWS and Google Cloud.

From Stargate to a $500 billion financing plan, Nvidia is building a full-chain AI infrastructure empire covering chip sales, project financing, equity investment, and debt guarantees.

“Circular transaction” controversy: $750 billion “left hand over right hand”?

However, the decline in Nvidia's stock price after the announcement of this financing plan is closely related to the market's continuing concerns about so-called “circular transactions.” Nvidia's financing empire is facing an increasingly sharp question: when the supplier is both the customer's investor and financial guarantor, how much is the actual terminal demand?

Critics warn that such deals are artificially boosting demand and valuations across the AI industry. The core mechanism is that Nvidia provides investment or financing guarantees to AI companies (such as OpenAI and Anthropic). These funds are then used to purchase Nvidia's GPUs, forming a closed loop of capital, thus creating the illusion of natural demand and strong revenue streams.

This chain is more complicated than it seems. Nvidia invests in basic AI companies such as OpenAI and Anthropic, and these AI developers are also Nvidia's biggest customers. The capital network even extends to cloud service providers such as Oracle — Oracle is Nvidia's chip customer, OpenAI is Oracle's computing customer, and Nvidia is also an investor in OpenAI. The same amount of money is calculated over and over again in the same ecosystem, creating the illusion that demand is being amplified multiple times.

Jim Cramer bluntly warned: “I've been through 2000 and I don't want a sequel.” The “big short” investor Michael Burry further warned that Nvidia's “excessive expansion” is driving revolving spending to a “biblical scale.”

The credit market has cast its first vote of no confidence. Nvidia's 5-year debt default protection costs once saw the biggest one-day increase since the contract began trading in November 2025.

Echoes of History: Warning of the Internet Bubble

The current AI boom has one key thing in common with the 2000 internet bubble: circular transactions have boosted infrastructure demand. What was fiber optic cable in the 90s is now GPU and data center capacity.

The bursting of the internet bubble erased $5 trillion in market capitalization. It took 8 years for Amazon to return to its peak before the crash, and 25 years for Cisco. But the key difference today is the size of the participants and the high degree of systemic integration — tech giants such as Amazon, Microsoft, Google, and Apple are deeply rooted in the global economy, and their shares are widely held in 401 (k) retirement plans and index funds. Once there is a systemic adjustment, the consequences may far exceed the 2008 financial crisis.

In the late 90s of the last century, telecom equipment vendors such as Nortel, Lucent, and Cisco also provided large-scale loans to startups to help them purchase their own routers and optical transmission equipment. Before the bubble burst in 2001, the total financing provided by the top five telecom equipment companies exceeded 123% of the total net profit. Bespoke Investment Group expressed the same concerns as back then: “The entire AI industry seems to circulate itself. If Nvidia can only grow by providing capital and turning it into revenue, then it is questionable whether the entire ecosystem is sustainable.”

Nvidia, on the other hand, has adopted a very different strategy from Cisco's. Cisco “only sells equipment” during the Internet bubble, and Nvidia is using a financing platform to position itself at the core hub of AI capital flows — both a “seller,” a “lender,” and an “investor.” This tripartite role allows it to reap maximum benefits in upward cycles, but it also means facing the greatest risk exposure in downward cycles.

If demand for artificial intelligence products suddenly weakens, or profit times exceed investors' expectations, the ensuing adjustments could cause far-reaching economic shocks. Currently, the entire industry is playing an “ultimate gamble,” betting that the potential utility of artificial intelligence will eventually verify its astronomical valuations and the unprecedented infrastructure construction that supports these valuations.