Nvidia’s new financial strategy does not compute
Get the Finance newsletter
Daily finance — markets, central banks, M&A, the prints that move money. Free.
- Nvidia is working with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR on a $500 billion financing consortium aimed at establishing compute as an investable asset class
- Jensen Huang now describes Nvidia chips as "revenue-generating," "long-lived," and "fungible" assets — a sharp reversal from his 2024 claim that Hopper GPUs would be unsellable once Blackwell shipped in volume
- BlackRock CEO Larry Fink compared the moment to "the very beginning" of mortgage-backed securities in the 1970s, calling compute "a next future for financial engineering"
- The consortium deals remain memorandums of understanding only; Nvidia's similar $100 billion MoU with OpenAI from 2024 never materialized into a binding transaction
- Huang claims the 2020-era A100 chip has an economic life "toward a decade," a figure that directly expands the borrowing capacity of GPU-backed loan recipients like CoreWeave, whose loan limits track chip depreciation schedules
- Broadcom executed a comparable $35 billion deal with Apollo and Blackstone earlier in 2025, using roughly one million chips as collateral for senior notes guaranteed by Broadcom
- CME Group plans to launch compute futures in October, pending regulatory approval of two contracts under review
Why it matters: Huang's 10-year depreciation claim for older chips directly inflates how much banks can lend against GPU collateral, reshaping loan terms across the industry. The deals are MOUs only — Nvidia's $100B OpenAI pact from 2024 never closed — but the "compute as asset class" framing still gives Nvidia's biggest customers a longer amortization runway regardless of whether the consortium ever funds a dollar.
Ask SkimNews
