Nvidia has struck partnerships with six of the world's largest financial institutions, Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR, to raise more than $500bn (roughly €460bn) for the infrastructure underpinning the artificial intelligence boom. The chipmaker announced the deal on Monday, saying the money would fund data centres, chip factories and the power stations needed to run them.
According to the Guardian, which first reported the news for a European audience, chief executive Jensen Huang called the arrangement "a major milestone for Nvidia and the AI industry" in a post on X, alongside a photograph of himself with the finance chiefs involved, including Goldman Sachs boss David Solomon.
Turning chips into an asset class
Nvidia's own statement, carried by its newsroom, said the new platforms are designed to be independent structures that mobilise capital from pension funds, insurers and other investors rather than sitting on Nvidia's own balance sheet. The company framed the move as a response to surging global demand, noting that governments, enterprises and cloud providers are racing to secure computing capacity.
Goldman Sachs' Solomon described his firm's role as building "a market for credit backed by NVIDIA compute", language that suggests the platforms could eventually package and sell debt linked to chip-based infrastructure, similar to how banks securitise mortgages or car loans.
“"We're in a pivotal moment of a historic AI investment cycle," Solomon said, according to CNBC.”
Circular financing worries resurface
The scale of the deal has reignited a debate that has followed Nvidia for much of the past year: whether the AI industry's financing has become dangerously circular. Axios reported that the move comes weeks after reports Nvidia was in talks to guarantee financing for a quarter-trillion-dollar data centre project tied to OpenAI, one of its biggest customers, raising questions about how much of the new capital ultimately flows back to Nvidia itself through chip purchases.
Bloomberg has previously reported on a related $500bn-plus agreement between Nvidia and South Korea's SK Group, where Huang said the value included Nvidia's own purchases of memory chips as well as SK's purchases of Nvidia supercomputers. Critics argue that when a supplier finances, or helps finance, the very companies buying its products, it becomes harder for investors to judge genuine demand from demand manufactured by the financing arrangement itself.
Huang has pushed back on that framing, arguing that graphics processing units generate real, ongoing cash flow for their owners and should be treated as productive assets rather than depreciating equipment, according to Bloomberg.
What it means for Europe
For European readers, the deal matters beyond Wall Street. Nvidia has been directly involved in financing AI buildouts on the continent, including France's Mistral AI, which raised €722m in debt financing to build Nvidia-powered data centres, starting with a facility near Paris. Nvidia has said Europe needs sustained investment in power supply and digital infrastructure to keep pace with the US and Asia in AI development.
With European governments already grappling with high energy costs and slower growth than the US, cheaper, large-scale financing for data centres could accelerate the continent's AI ambitions. But it also means European utilities, cloud operators and eventually taxpayers could be more exposed if the debt-fuelled AI investment cycle Nvidia is helping to build does not deliver the returns investors expect.
The six firms involved in the new platforms manage trillions of dollars combined, and their executives, including BlackRock's Larry Fink and Blackstone's Jon Gray, have publicly backed the strategy. Whether the arrangement proves to be sound infrastructure investment or a fragile edifice built on interlocking corporate promises is likely to remain a central question for markets watching the AI boom through 2026.
This article is free to read. It always will be — no paywall, no account, no tracking.
