As of the past few weeks, debt markets have almost entirely shut off access to money for any company that doesn’t have substantial revenue independent of AI data centers.
SpaceX closed $40B of debt for GPUs, but we’re investment grade and have Starlink & launch revenue.
That means the 200+ neoclouds are going to see guppy-level death and consolidation. Some will make it through, but not many.
The crux of the issue is that the financing clock of AI data centers starts when the chips arrive but the energized shell clock starts years earlier and often finishes later and the cash to service the debt starts only once the AI hardware is actually running.
An energized shell is a data-center hall with power and delivery gear in place, so installed compute can run. Land, interconnection, transformers, switchgear, and commissioning take years. A new electrician who starts an apprenticeship in late 2026 finishes in 2030 or 2031 /QZ/. The AI hardware does not wait on that crew and often sits unpowered in warehouses around the world.
Under a new Early Rack Receipt process, AWS expected to receive roughly $5 billion in Nvidia’s GB300 server racks at US facilities before final deployment, according to August internal emails reviewed by Business Insider. The company projected roughly $3 billion more in GB300 racks each month “for the foreseeable future,” one of the emails said /BI/.
That is the mismatch. A signed GPU order and a financing vehicle can both come due while the building is still dark. The lender is paid from revenue the idle machines are not yet earning. In the last week of October, debt for a pure hall stopped clearing. The paper that still moved belonged to borrowers with cash flow besides the hall, such as SpaceXAI’s Starlink and launch.
So let’s look deeper at the operating economics underneath this timing mismatch in Constraint Capital #10...