Nvidia (NVDA.US) brings in Wall Street financial giants such as Goldman Sachs and Blackstone to plan a $500 billion AI financing plan to buy “blood transfusions” for customers to buy chips

The Zhitong Finance App learned that Nvidia (NVDA.US) is collaborating with major Wall Street financial institutions such as Goldman Sachs (GS.US), Blackstone Group (BX.US), and Apollo Global Management (APO.US) to plan AI computing power financing plans with a scale of up to 500 billion US dollars, hoping to help customers such as OpenAI and Anthropic obtain the huge capital needed to purchase Nvidia chips, thus providing new financing channels for future AI infrastructure expansion.

According to people familiar with the matter, Goldman Sachs, Blackstone, and Apollo have been designing related debt financing plans for the past few months, but progress has been relatively slow due to the complicated transaction structure. Subsequently, Nvidia CEO Hwang In-hoon decided to make this plan public and announced that relevant financial institutions plan to jointly finance AI computing power projects with a total scale of about 500 billion US dollars.

Notably, $500 billion is not an amount of financing that has been signed or formally committed. According to people familiar with the matter, the figure includes both potential transactions currently being discussed and predictions of future financing needs. There is currently no clear deadline for completion. As of the announcement of the plan, the relevant parties had not formally signed a specific financing transaction.

In addition to Goldman Sachs, Blackstone, and Apollo, Nvidia also notified three companies a few days before the plan was announced that KKR (KKR.US), BlackRock (BLK.US), and Brookfield (BAM.US) will also join the financing camp. This means that several of the world's largest banks, private equity, and asset managers will participate in this financing system.

For Nvidia, one of the important purposes of the program is to prove to investors that its AI chip customers have sufficient financing support behind them. Nvidia's future demand depends not only on large cloud computing companies such as Microsoft (MSFT.US) and Amazon (AMZN.US), but also increasingly on AI startups such as OpenAI and Anthropic. Compared with large technology companies, these AI companies themselves have relatively limited cash flow and financing capacity, so how to fund huge chip purchases and computing power construction has become an important issue for whether the AI investment cycle can continue.

Nvidia's previous practice of investing in its own customers such as CoreWeave (CRWV.US) raised market concerns about “circular financing,” that is, chip companies invest in customers, and customers then use funds to buy chips, which may artificially amplify AI demand. The $500 billion financing plan initially also caused some bond investors to worry that Nvidia might take too much leverage risk.

Subsequently, Hwang In-hoon further clarified that the support provided by Nvidia will only cover up to 25% of individual financing opportunities, and that the company will evaluate each project instead of providing a unified guarantee for the entire $500 billion financing. This explanation partly allayed market concerns about Nvidia's potential risk exposure.

People familiar with the matter said that financial institutions participating in the plan are currently in contact with customers such as sovereign wealth funds, pensions, and insurance companies to understand their intention to buy related debts. Some of the funds may even come from individual investors in the future.

Judging from the financing structure, most of the capital is expected to come from the private credit market, but considering the potential size of 500 billion US dollars, it is difficult to rely on private capital alone to meet demand, so the open bond market will also become an important source of capital. Some of the transactions may issue tens of billions of dollars of bonds through special purpose carriers. These carriers will buy Nvidia chips and then rent the chips to AI customers such as OpenAI and Anthropic.

The collateral assets for the loan in question are expected to include the chip itself and the long-term purchase or use agreement signed by the customer. If an AI customer is ultimately unable to bear the cost of the chip, the relevant GPU can theoretically still be leased to other companies, thereby reducing the losses caused to creditors by a single customer default.

However, there are still risks in this financing model. Currently, Nvidia's high-end GPUs are highly valuable due to strong AI demand. If AI infrastructure construction expands excessively in the future and causes an oversupply of computing power, the second-hand value and rental price of the chip may drop, thereby weakening the value as collateral for debt.

Nvidia isn't the only company trying to use Wall Street capital to support AI customers to buy chips. Broadcom (AVGO.US) also announced a partnership with Apollo and Kuroishi a few weeks ago, and plans to finance the construction of more than 20 gigawatts of computing power capacity for cutting-edge AI laboratories, including Anthropic and OpenAI, by 2028, with potential capital requirements also reaching hundreds of billions of dollars. Unlike Nvidia, Broadcom had already received $35 billion in financing from Apollo and Blackstone when it announced the partnership.

As demand for AI infrastructure capital continues to expand, more Wall Street institutions also want to enter this market. According to reports, J.P. Morgan Chase (JPM.US) is discussing ways to participate in Nvidia's related financing plans. Morgan Stanley (MS.US) launched a $1.5 trillion financing framework shortly after Nvidia announced plans to support US innovation and national security-related investments, with AI and advanced computing listed as key areas.

Judging from the current situation, Nvidia's $500 billion plan is more like building a huge AI financing platform than an already implemented $500 billion funding commitment. Its core goal is to connect Nvidia's growing demand for chips with Wall Street, private credit, pensions, insurance funds, and open bond markets to provide financing support for AI companies to continue to expand computing power, while reducing the balance sheet pressure brought about by Nvidia's own direct provision of capital to customers.