Jensen Huang announced on Tuesday that Nvidia is working with a consortium of Wall Street firms, including Apollo Global Management, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR, to set up financing vehicles aimed at raising more than $500 billion for AI compute infrastructure.
What the deal looks like
The plan is to pool money from "third-party investors" such as pension schemes and insurers, allowing AI-focused companies to lease Nvidia GPUs and data-centre capacity rather than buying the hardware outright. By keeping the risk off Nvidia's balance sheet, the chipmaker can sell its products while the financing arm bears the debt.
"We have moved from an era in which companies bought chips and built data centres project by project to one in which AI factories can be financed as productive infrastructure," Huang wrote. "In AI, compute is revenue."
Why it matters for Europe
European insurers and pension funds are increasingly looking for long-duration assets that match their future liabilities. Data centres, with lifespans of decades, fit that profile. By turning compute into an asset class similar to toll roads or power plants, Nvidia is opening a new source of capital that could flow into European data-centre projects, potentially lowering financing costs for local AI start-ups and cloud providers.
Analysts have warned that without such structures, GPUs could be seen as rapidly depreciating equipment, requiring ever-greater cash injections. The new financing model treats the hardware as a revenue-generating asset, making it more attractive to debt markets.
What comes next
Details of each partnership remain scarce, but the consortium is expected to launch the first financing vehicles within weeks. Nvidia has hinted it may provide residual-value support of up to 25 % for some projects, offering a safety net if the underlying chips lose value.
Wall Street analysts such as Joseph Moore of Morgan Stanley view the approach as a way to allay concerns about circular financing, while Vivek Arya of Bank of America notes that Nvidia's GPUs are especially financeable because they can be redeployed across operators and extended with CUDA software.
Critics, including technology strategist Ben Thompson, caution that pension and insurance capital is fundamentally risk-averse, unlike venture funding. The shift could therefore impose new constraints on how AI projects are funded, blending the high-growth tech sector with the safety-seeking world of long-term investors.
For now, the success of the initiative will hinge on whether investors are comfortable treating compute as a stable, income-producing asset. If they are, the model could reshape the financing of AI across the continent, providing a steady stream of capital that matches the long-term nature of data-centre investments.