Nvidia partners with Apollo, BlackRock and others on AI financing push targeting over $500 billion

Nvidia partners with Apollo, BlackRock and others on AI financing push targeting over $500 billion

Huang is pitching AI computing as infrastructure-like, revenue-producing capacity that could be financed more like power plants or toll roads than a short-lived technology purchase.

Fact Check
Nvidia's own investor relations press release confirms the $500B+ SK Group partnership and SK Telecom's planned 2-gigawatt AI factory, validating the first part of the claim directly from a primary official source. The OpenAI Ohio financing is accurately framed as 'reported' — the WSJ originated the report and it was widely corroborated (Quartz, NYT, Yahoo Finance), describing Nvidia's ~$250B backstop for OpenAI's ~10GW Ohio project. The FT report on Wall Street/Nvidia AI financing supports the framing that Nvidia is moving beyond chip supply into backing data center buildouts. All claim elements are supported.
Summary

Nvidia and six major Wall Street firms are sketching out financing platforms aimed at mobilizing more than $500 billion of mostly third-party capital over time for AI infrastructure, as Jensen Huang argues that AI compute should be financed like long-lived, revenue-producing infrastructure rather than treated as a fast-depreciating technology expense. The arrangements with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR remain preliminary and are governed by memorandums of understanding rather than binding contracts, with borrowers, rates, locations and deployment timelines still largely unspecified. Huang said the push would help customers build out AI data centers and lock in demand for Nvidia hardware by reframing specialized computing capacity as a bankable asset that can generate income over years and serve multiple users. He described Nvidia's chips as part of infrastructure similar to electricity or the internet, while the company says its so-called AI factories use power and data to produce intelligent systems for applications including chatbots, image and video generation, drug design and robotics. The effort follows an initial AI buildout funded largely through debt and equity issuance by major technology companies, with investor debate still centered on depreciation, utilization, residual values and the risk of excess capacity.

Terms & Concepts
  • AI compute: The processing power used to train and run artificial intelligence models, typically provided by specialized chips in data centers.
  • AI infrastructure: The physical and software systems, including chips, servers, networking and data centers, that support large-scale AI workloads.
  • AI factories: Nvidia's term for data-center systems that turn computing power, electricity and data into AI services and applications.