BlackRock, Goldman, Apollo, Blackstone, Brookfield and KKR are in talks with Nvidia on an AI buildout that could reach $500 billion

BlackRock (NYSE: BLK), Goldman Sachs (NYSE: GS), Apollo Global Management (NYSE: APO), Blackstone (NYSE: BX), Brookfield Asset Management (NYSE: BAM) and KKR (NYSE: KKR) are lining up with Nvidia (NASDAQ: NVDA) for a giant new round of AI spending that could eventually reach $500 billion.

According to the Financial Times, the collaboration is expected to revolve around funding additional computer capabilities, electricity provision, and constructing data centers. The deal can be unveiled as early as next Monday.

NVDA fell by 1.4% following the news, thereby resulting in a loss of over $70 billion in the company’s valuation.

The discussions come as Nvidia takes a bigger role in paying for the huge physical buildout behind artificial intelligence.

The company is worth about $5.25 trillion and already sells the graphics processors used to train and run most of the biggest US AI systems. It also supplies software and other computing tools. But GPUs alone do not keep an AI model running.

For operators, it requires huge amounts of server-containing buildings, power, cooling, and financing over an extended period of time. Nvidia has been contributing increasingly to helping its clients get this funding, along with putting their funds in the companies buying its hardware.

Nvidia backs more AI customers while Mark Cuban warns the financing could “crumble”

Billionaire Mark Cuban has been openly worried about how much debt and financial engineering now sit behind the AI spending boom.

His argument is simple. Nvidia is no longer just collecting money when somebody orders chips. It has also been helping finance the companies placing those orders.

In July, Mark compared the current setup with the funding frenzy that surrounded internet companies during the dot-com period. He wrote that “instead of IPOs, Nvidia is the IPO, funding everyone and anyone.”

That funding can take different forms. Nvidia has offered capital, revenue-sharing deals and arrangements that guarantee certain levels of income to data center operators and newer cloud providers. Those deals can make it easier for customers to order GPUs and start building before their own businesses produce enough cash to cover the full cost.

During the first half of 2026, Nvidia had invested more than $40 billion in its AI endeavors. This amount included investments and funds related to companies such as OpenAI, Corning (NYSE: GLW), and IREN (NASDAQ: IREN).

Some of the numbers are massive even by AI standards. OpenAI has been linked to a proposed $100 billion data center program. Financing tied to xAI has also used special-purpose vehicles worth billions of dollars.

These arrangements may rely on pledges of GPUs as collateral for loans. In addition, they may involve leases of long duration and projections of revenues based on emerging AI companies.

This is relevant in case the true earnings do not meet the figures that the financiers relied upon at the time of signing of the contracts.

It has already been noted by legal experts that AI infrastructure financing has become quite complex. One single project may comprise private lending, securitization of debts, and SPVs formed specifically for the purpose of holding assets or borrowing without a balance sheet.

This may involve banks, insurance companies, and pension funds.

Wall Street puts more capital behind GPUs as hardware values fall faster

The value of the hardware itself is also a danger.

For instance, Nvidia has been rolling out new generations of AI chips once every year. That implies that a data center might still be in the process of paying off one batch of processors when an even better version becomes available.

As a result, old GPUs may depreciate faster than was assumed by lenders in the beginning.

This point is important if we take into account the fact that it is the hardware itself that guarantees the loan. For example, banks or private credit facilities may give loans for equipment that has very high costs at the moment of installation.

A slowdown in spending on AI might damage operators who invested in capacity to meet demand that did not materialize. More frequent hardware updates might make costly installations obsolete sooner. An improved chip from a rival might cut demand for Nvidia-based systems.

Either one of these factors might result in operators having to repay large debts against facilities generating lower revenue than expected.

It is much more than just investing in normal technology stocks. Leveraged finance is being used to finance data centers, GPUs, and energy projects.