Nvidia has brought together Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to create financing platforms capable of mobilising more than $500 billion for companies building artificial intelligence infrastructure.
The move marks a broadening of Nvidia’s role beyond chipmaking, as the world’s most valuable company seeks to help fund the factories, data centres, power infrastructure and advanced equipment required for the next stage of the AI industry.
It also reflects a growing shift in the technology race, with access to capital becoming as important as advances in models and processors. Companies and countries able to raise money cheaply and at scale are likely to have a significant advantage.
The five largest US hyperscalers are expected to spend $730 billion on capital expenditure in 2026. Up to a third of that spending is forecast to be financed through debt, rather than funded entirely from their own balance sheets.
Project finance has traditionally been used for infrastructure such as power stations, toll roads and airports, where large initial costs can be repaid over many years. Nvidia’s initiative seeks to apply a similar approach to computing, potentially making it an asset class suitable for pension funds and insurers.
The comparison has limits. Energy facilities are generally built to operate for decades and generate relatively predictable returns, while graphics processing units have shorter lives and demand for AI capacity remains difficult to forecast.
Even so, the proposed structure could give a wider group of companies access to funding beyond the major hyperscalers, which can meet only part of the cost of the build-out themselves.
The arrangement has prompted concerns about circular financing, in which money provided by Nvidia could return to the chipmaker through orders for its products. However, Nvidia is not expected to lend most of the money itself, and neither are its six financial partners.
Instead, the capital is intended to come from third parties including pension funds, insurers, sovereign wealth funds and household savings. That would connect the expansion of AI infrastructure to the wider financial system rather than creating a closed funding loop.
Capital markets become central to the AI race
The financing challenge is particularly significant outside the United States. Europe has a large pool of savings, with eurozone households holding about 10 trillion euros in bank deposits, but institutional investors often prefer US technology bonds to backing European start-ups.
That leaves Europe with money available but without the financial mechanisms needed to direct enough of it into technology investment. A stronger capital-markets strategy, including progress towards greater integration, could therefore be central to the continent’s AI ambitions.
China faces a different difficulty. State-backed funding can support rapid construction, but may be less effective at matching new capacity with demand. Chinese media reported last year that as much as 80% of newly built computing capacity was unused after a policy-driven expansion.
The Stanford HAI AI Index 2026 estimates that private investment in AI in the United States was more than 23 times higher than in China last year.
The United States therefore begins the next phase with a financial advantage, supported by deep bond markets, private-credit funds and large asset managers capable of investing in risky, long-term projects.
The first stage of the AI boom was driven by models, chips and applications. Nvidia’s financing initiative suggests that the next stage will also depend on whether the financial system can support expansion beyond a relatively small group of dominant companies.
