Nvidia Taps Wall Street to Mobilize $500 Billion-Plus for AI Compute

Nvidia is teaming up with six of the world’s largest investment and financial firms on a new effort to mobilize more than $500 billion of third-party capital for artificial intelligence infrastructure, a move that could push AI computing further into the territory traditionally occupied by infrastructure and private-credit investing. The strategy is built around Nvidia’s argument that large-scale compute can be financed as a productive, investable infrastructure asset.

The chipmaker said it signed memorandums of understanding with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to establish independent compute-financing platforms. According to Nvidia’s announcement, the platforms are intended to create large pools of capital for customers building AI infrastructure across frontier AI labs, enterprises and cloud providers.

The $500 billion-plus figure is not a committed fund sitting ready for deployment. Nvidia described it as capital the partnerships intend to mobilize over time, and the company did not disclose how much each financial firm might contribute, what financing terms customers would receive or when the capital would be deployed. The partnerships are also subject to final agreements, making the announcement a framework for a potentially enormous financing market rather than a completed $500 billion transaction.

Nvidia is trying to make compute financeable like infrastructure

The most important part of the plan may be the financing model rather than the headline number. Nvidia is arguing that advanced computing capacity can be underwritten as a productive infrastructure asset, with investors funding AI systems in exchange for long-duration returns linked to their use. For Nvidia’s customers, the appeal is access to capital for costly data centers and computing systems without relying entirely on their own balance sheets for the upfront buildout.

Nvidia calls these facilities “AI factories,” reflecting its view that computing capacity is increasingly an economic input that can generate revenue as customers train and run AI models. The company said its platform is suitable for this approach because the hardware can serve different models, workloads, customers and operators, while its CUDA software ecosystem can extend the usefulness of the systems. Those are Nvidia’s claims about the economics of its platform, not guarantees of investment returns.

The financial firms bring a different set of capabilities. Apollo, Blackstone, Brookfield and KKR are major investors in private credit and infrastructure, while BlackRock and Goldman Sachs can connect large pools of institutional capital with debt and equity markets. Goldman Sachs CEO David Solomon said in Nvidia’s release that the initiative could help create a market for credit backed by Nvidia compute. That is a notable step beyond simply financing construction of data-center buildings, because it treats the computing equipment and the revenue generated from using it as part of the investable structure.

There is already a precedent for institutional capital moving toward AI infrastructure. In 2024, BlackRock, Global Infrastructure Partners, Microsoft and MGX launched an AI infrastructure partnership that initially sought $30 billion of private equity capital and as much as $100 billion of total investment potential including debt. Nvidia supported that initiative with technical expertise and later joined the partnership. The new Nvidia-led effort is broader in stated scale and involves multiple independent financing platforms rather than a single pool.

The structure could reduce the burden on AI builders, but the risk does not disappear

AI infrastructure requires enormous upfront spending on chips, networking equipment, data-center construction, power and cooling. Moving more of that cost into financing vehicles backed by asset managers and private-capital firms could allow AI developers and cloud operators to build capacity faster than they could through retained earnings or conventional corporate borrowing alone.

That does not make the economics risk-free. The investment case depends on sustained demand for computing capacity, the useful life and resale value of expensive hardware, the credit quality of customers and the terms under which capacity is leased or contracted. Rapid improvements in AI chips can also create an unusual underwriting problem: lenders and investors may be financing assets that are extremely productive today but face technological obsolescence faster than traditional infrastructure such as power plants, pipelines or office buildings.

Nvidia’s role may also go beyond supplying hardware. Reuters reported, citing a post by CEO Jensen Huang on X, that Nvidia has the option to backstop as much as $125 billion, or 25% of potential deals. That should not be read as a $125 billion funding commitment. Nvidia’s formal announcement did not disclose a backstop amount, and the exact terms of any support would depend on the final agreements and individual transactions.

The distinction matters because the initiative is designed around independent underwriting by the financial partners. If those firms ultimately provide most of the capital and make their own credit decisions, Nvidia can help expand the market for its computing platform without carrying the full financing burden itself. At the same time, any guarantees, backstops or other forms of support could leave Nvidia with financial exposure if demand or customer economics weaken. The company has not yet disclosed enough detail to quantify that risk.

A bigger financing ecosystem could reinforce Nvidia’s data-center business

The financing push arrives when data-center sales already dominate Nvidia’s financial results. In its fiscal first quarter of 2027, Nvidia reported $81.6 billion of total revenue and $75.2 billion of Data Center revenue. Data Center revenue rose 92% from a year earlier, underscoring how closely the company’s growth is tied to continued investment in large-scale computing infrastructure.

Creating additional sources of capital for customers could therefore have a direct strategic benefit. More financing capacity can potentially support more AI factories, which in turn can increase demand for Nvidia GPUs, networking products and software. The company’s announcement explicitly links the financing platforms to growth across both hardware sales and software adoption.

The strategy also broadens the group of investors that could gain exposure to the AI buildout. Until now, much of the most visible spending has come from the balance sheets of large technology companies and specialized AI infrastructure providers. A mature market for compute-backed credit or infrastructure vehicles could bring a wider pool of long-duration institutional capital into projects that previously looked more like technology-company capital expenditure.

Whether that happens at the scale Nvidia is targeting will depend on the details still missing. The company has not disclosed final agreements, individual capital commitments, pricing, leverage levels, customer contracts or a deployment schedule. Those terms will determine whether “compute as an investable asset” develops into a durable institutional market or remains a financing concept tied to the current surge in AI demand.

The next concrete step is the execution of final agreements with the six financial firms, followed by the structure and funding of individual platforms and transactions. Until those details emerge, the $500 billion-plus figure is best understood as Nvidia’s ambition for third-party capital mobilization, not money that has already been raised or committed.