Generative AI Fundamentals Map
Map/Loops/balancing

Payback and collateral

balancing loop · 4 steps · Trace on the map →

The cycle

Capex buys GPUs, oftenon debt secured by theGPUsSupply grows andrental prices fallOlder GPUs lose resaleand collateral valuePayback gap widens andlending tightensB
  1. 01Capex buys GPUs, often on debt secured by the GPUsDatacentre capex
  2. 02Supply grows and rental prices fall
  3. 03Older GPUs lose resale and collateral value
  4. 04Payback gap widens and lending tightensCapex payback gap

What drives it

The brake on the capital flywheel. Each wave of capex adds supply, which pushes rental prices and the resale value of older GPUs down. That widens the gap between spend and revenue and shrinks the collateral behind the next loan. The memory squeeze adds to it: commodity memory prices rose 90 to 95% in one quarter of 2026, raising the cost of every rack.

What breaks it

Demand outrunning supply. CoreWeave says expired H100 contracts rebooked at about 95% of the original price. Longer useful lives on the books delay the bite but do not remove it: a three-year-old H100 system resells for about 45% of the price of a new one.

What to watch

GPU rental and resale indices. Neocloud bond spreads. Useful-life changes in hyperscaler filings. The September 2026 selloff was this debate.

Also feeds it

Sources

  1. 01Value the Markets: the GPU depreciation debate behind this week’s AI dip (Sep 2026)
  2. 02Hyperscaler useful-life changes (Dow Jones via Tiger, Dec 2025)
  3. 03GPU financing filters into ABS (The Star/Bloomberg, 7 Oct 2026)
  4. 04Memory and NAND flash crisis, May 2026 update (NAND Research)
  5. 05CIO: $6T in annual AI revenue needed to pay off data centres (Bain)