Nvidia, the chip company led by Jensen Huang, is mobilizing up to $500 billion through a financing partnership with major investment funds, according to an announcement. The project brings together some of the world's largest asset managers, including Apollo Global, Blackstone, Global Infrastructure Partners of BlackRock, Brookfield Asset Management, Goldman Sachs, and KKR. The objective is to allow Nvidia's clients—cloud operators, data center builders, or large technology groups—to buy computing power without having to finance all of their investments alone.
Nvidia's $500 billion financing envelope is intended to finance buyers of graphics processing units (GPUs) who are building AI infrastructure. The company announced an agreement with major investment funds to sell GPUs to its clients on credit. This arrangement indirectly secures client demand and accelerates the construction of 'AI factories' (usines d'IA), according to reports.
Cloud giants are expected to invest $730 billion this year, largely in chips. These companies, including Google, Amazon, and Microsoft, will be able to borrow to buy chips, allowing them to increase their computing capacity faster and theoretically earn more money by renting servers, thereby repaying their credit.
John Plassard, head of investment strategy at Cité Gestion, estimated that the financing would go through dedicated vehicles that borrow on bond markets or from private investors before renting computing capacity to end users. Jacques Lemoisson, founder of Gate Capital Management, said of Nvidia's announcement: 'It's an acceleration of circular financing!'
$500 billion
The financing envelope Nvidia is mobilizing through its partnership with major investment funds.
Nvidia has not given many details about the financing arrangement so far. Despite the announcement, Nvidia's stock fell. Individual savers are already involved in Nvidia's $500 billion AI financing without knowing it, according to a report.
US stock exchanges defied uncertainties over oil prices, while Nvidia's financing announcement drew comparisons to the telecom bubble in some commentary, though other sources framed it neutrally.
Updates
Nvidia’s CEO Jensen Huang confirmed the company may backstop up to $125 billion — or 25% — of the $500 billion AI infrastructure financing plan, while new disclosures reveal talks with OpenAI to guarantee up to $250 billion for a major Ohio data center project; shares dropped over 3% following the report, and rental rates for H100 chips rose from $1.70 to $2.35 per GPU-hour, though disputes persist over whether seven — not six — financial firms are formally involved.
Nvidia’s CEO Jensen Huang confirmed the company may backstop up to $125 billion — or 25% — of the $500 billion AI infrastructure financing plan, while new disclosures reveal the firm is in talks to guarantee up to $250 billion for OpenAI’s Ohio data center project; shares dropped over 3% after the report, and rental rates for H100 chips rose from $1.70 to $2.35 per GPU-hour this year, though concerns persist over potential depreciation and circular financing risks, as noted by FedWatch’s Ben Emons and Bank of America.
Nvidia’s CEO Jensen Huang confirmed the company may backstop up to $125 billion — or 25% — of potential deals under the $500 billion AI infrastructure financing plan, while new disclosures reveal Nvidia is in talks to guarantee up to $250 billion for OpenAI’s Ohio data center project; additionally, rental rates for H100 chips rose from $1.70 to $2.35 per GPU-hour this year, and China’s impending travel restrictions on AI engineers, effective September 15, have raised concerns over supply chain risks, according to Ben Emons of FedWatch Advisors.
Nvidia confirmed it may backstop up to $125 billion — or 25% — of deals under its $500 billion AI infrastructure financing plan, while also revealing it is in talks to guarantee as much as $250 billion for OpenAI’s Ohio data center project; shares dropped as much as 3.2% following the report, and CEO Jensen Huang emphasized the company’s role in creating a new investable asset class, with rental rates for H100 chips rising from $1.70 to $2.35 per GPU-hour since late 2025.
Nvidia's CEO Jensen Huang confirmed the company may backstop up to $125 billion—25% of the $500 billion initiative—to support AI infrastructure deals, while new disclosures reveal Nvidia is in talks to guarantee up to $250 billion for OpenAI’s Ohio data center project; shares fell as much as 3.2% amid concerns over depreciation risks and China’s emerging AI chip competition, which could undermine collateral values in private credit markets.
Nvidia confirmed it has the option to backstop up to $125 billion — 25% of the $500 billion initiative — for AI infrastructure deals, while new details emerged that it is in talks to guarantee up to $250 billion for OpenAI’s Ohio data center project; additionally, rental rates for H100 chips rose from $1.70 to $2.35 per GPU-hour, and China’s planned travel restrictions on AI engineers, effective September 15, were reported as a potential threat to Nvidia’s financing model by analyst Ben Emons.
Nvidia disclosed it may backstop up to $125 billion — 25% of the $500 billion initiative — for AI infrastructure deals, while also confirming talks to guarantee $250 billion for OpenAI’s Ohio data center project; additionally, rental rates for H100 chips rose from $1.70 to $2.35 per GPU-hour, and China is set to impose travel restrictions on AI engineers at firms like Alibaba and DeepSeek starting September 15.
Nvidia confirmed it has the option to backstop up to $125 billion — 25% of the $500 billion initiative — exposing its balance sheet to contingent liabilities, while new disclosures reveal it is in talks to guarantee up to $250 billion for OpenAI’s Ohio data center project, a figure exceeding earlier estimates; additionally, rental rates for H100 chips rose from $1.70 to $2.35 per GPU-hour this year, and China’s planned travel restrictions on AI engineers, effective September 15, introduce new geopolitical risk to the financing model.
Nvidia confirmed it has the option to backstop up to $125 billion — 25% of the $500 billion initiative — exposing its balance sheet to contingent liabilities, while new disclosures reveal talks to guarantee up to $250 billion for OpenAI’s Ohio data center and that rental rates for H100 chips rose from $1.70 to $2.35 per GPU-hour; additionally, China is set to restrict AI engineer travel from key firms starting September 15, and the US FCC is considering a ban on China-sourced optical transceivers, heightening supply chain risks.
Nvidia has confirmed it may backstop up to $125 billion — 25% of the $500 billion initiative — exposing its balance sheet to contingent liabilities, while also revealing new talks to guarantee up to $250 billion for OpenAI’s Ohio data center project; additionally, rental rates for H100 chips rose from $1.70 to $2.35 per GPU-hour, and China’s planned travel restrictions on AI engineers, effective September 15, have introduced new geopolitical risk to the financing model, according to FedWatch Advisors.
