CME's planned Oct. 5 launch of Nvidia GPU rental-rate futures now sits alongside a broader debate over whether AI chips can retain enough collateral value to underpin large-scale financing. Financial Times and other media reported on Aug. 13 that Nvidia had signed memorandums of understanding three days earlier with Apollo Global Management, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs and KKR to create dedicated funding pools for AI infrastructure, part of an effort described as mobilizing more than $500 billion largely from third-party capital. The model depends on whether GPUs can keep generating cash and hold their value as newer chips arrive: Jensen Huang said H100 chips are retaining value longer than expected, rental rates have recently risen, and A100 chips launched six years ago are still being used beyond their originally expected lifespan. Lenders nevertheless often require full principal repayment within three to five years because rapid hardware obsolescence, weaker demand, overproduction or more efficient AI models could erode resale values. Supporters argue that GPU leasing, price benchmarks, futures and selective residual-value support could help turn compute into a financeable asset class, while critics warn that fast-changing semiconductors are harder to value than traditional leased assets such as cars or aircraft.