All the books

Book · 3 contracts · 4 expiries

The debt behind the machines

The hyperscalers' lease obligations, what a used GPU fetches as collateral and the term premium of the money underneath.

The theses it settles

The contracts

The next expiry of each; a strike opens its question, the title its whole ladder.

Who sits on each side

The commercial parties a print hurts, on each side: the ones who would hedge with a contract on it.

  • Operating lease liabilities of the four hyperscalers, growth from a year earlier
    Loses if it rises
    • hyperscaler creditors who rank beside the lessors
    • rating agencies' adjusted leverage
    Loses if it falls
    • data-center developers and REITs that lease to them
    • the securitization buyers of their rent
  • GPU resale value (used data-center GPUs)
    Loses if it rises
    • buyers of used GPU fleets (second-tier clouds, universities, enterprises)
    • lessees with buyout options at fair value
    Loses if it falls
    • lenders and lessors holding GPUs as collateral or residual value
    • insurers writing GPU residual value cover
  • 10-year Treasury term premium (Kim-Wright)
    Loses if it rises
    • issuers rolling AI capex debt, such as hyperscaler bond programs and data-center ABS sponsors
    • holders of long-duration bonds and data-center real estate valued on long cash flows
    Loses if it falls
    • borrowers who locked fixed rates on data-center construction loans through swaps
    • money funds and short-duration lenders whose relative carry shrinks