Nvidia has partnered with six major financial firms to mobilize over $500 billion in third-party capital for AI infrastructure projects, including data centers, chip factories, and power plants. The company is guaranteeing up to 25% of the residual value of its chips installed in these projects to secure the financing. This move aims to address the challenge many AI companies face in accessing large-scale capital for compute needs, according to the Financial Times, which reported the deal. Nvidia's stock dropped about 1.4 percent following the announcement, wiping out more than $70 billion in market cap. The company's CEO, Jensen Huang, described the initiative as a shift from one-off projects to repeatable financing platforms, positioning AI infrastructure as a productive asset akin to power grids or transportation networks. Huang emphasized that the $500 billion target is an aggregate spread over years, not a single fund or a commitment to any one customer. He also noted that the actual credit assessment remains with the capital providers, with Nvidia only offering residual-value guarantees on a project-by-project basis. Source: thedecoder
Nvidia is taking on some of the depreciation risk for its chips by covering part of the gap if the resale or reuse value of installed hardware falls below expectations. The company claims this share is 'significantly lower' than in other compute financing arrangements. Huang also addressed criticisms that its financing of neoclouds and AI companies is circular, arguing that the guarantees are for individual projects rather than overarching commitments. He pointed to the long economic lifespan of its A100 chip, launched in 2020, which is still in commercial use six years later. Huang argued that CUDA keeps improving installed hardware over time, and that rising rental prices support the value proposition. For instance, H100 annual contracts rose from $1.70 per GPU-hour in October 2025 to $2.35 in March 2026, while B200 capacity runs between $5.30 and $7.05. Source: thedecoder
Critics like investor Michael Burry have warned that the rapid obsolescence of GPUs due to Nvidia's two-to-three-year upgrade cycle could lead to significant understatements in depreciation, with an estimated $176 billion loss between 2026 and 2028. Huang countered by asserting that the A100's economic lifespan extends toward a decade, citing the rising rental prices as market evidence. The Bank of England's July Financial Stability Report warned that the pace of AI infrastructure investment is historically unprecedented and that a shock to highly leveraged AI companies could trigger a global credit crunch. Source: thedecoder
