Nvidia’s new $500B plan is risky but brilliant, especially for aging GPUs
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Nvidia announced that Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR are willing to commit up to $500 billion to build AI data centers, a figure that spooked bond markets enough for CEO Jensen Huang to publicly defend the structure.
- Nvidia agreed to guarantee its own chips' value when used as collateral, pledging to cover up to 25% of the difference if lenders must liquidate and the GPUs cannot command the price the books assumed.
- Jensen Huang wrote on X that the structure is meant to address circular-financing concerns by bringing 'independent, long-term institutional capital' into AI infrastructure, explicitly distinguishing the plan from Lucent-era telecom lending.
- Nvidia has committed billions to frontier labs OpenAI and Anthropic and to neoclouds including CoreWeave, Nebius, Firmus, and Lambda, while Bloomberg calculated roughly $750 billion in additional circular deals this summer.
- Huang positioned Nvidia's servers as 'AI factories' — durable infrastructure akin to railroads or airlines rather than fast-depreciating assets, arguing a broad ecosystem of secondary users will protect residual chip value.
- Hyperscalers have leaned on debt (Oracle), new equity (Google), and cash burn (Meta) to fund their AI buildouts; Microsoft CEO Satya Nadella recently recommended the book '1873,' which chronicles the railroad boom's financial collapse.
- Nvidia's guarantee creates 'wrong way' risk: its obligations grow precisely as AI demand weakens, threatening to squeeze revenues and chip values at the same moment.
Why it matters: Nvidia is putting its own balance sheet behind a $500 billion AI infrastructure bet, agreeing to backstop up to 25% of any value drop on its GPUs used as loan collateral. That converts Nvidia from a chip supplier into a financial guarantor whose downside worsens exactly when AI demand cools — and it does so just as hyperscalers (Oracle on debt, Google on equity, Meta on cash burn) are visibly exhausting traditional funding channels.
Ask SkimNews
