Premium: The Hater's Guide To AI Debt (Part 1)

Newsletter · Watch/Read · (gathered 2026-09-23)

• The core argument is that the AI buildout is creating an “AI Debt” bubble, with major hyperscalers like Google, Oracle, and Amazon becoming cashflow negative due to massive capital expenditures on expensive AI chips and data centers, while AI services generate little revenue.
• The article introduces “neoclouds” – new companies, often converted cryptocurrency miners, that raise significant debt to build AI data centers and acquire chips, with their capital expenditures far exceeding their current revenues.
• Over $500 billion in AI-related debt was issued in 2026 alone, funded by private credit, investment banks, and global bond markets, with estimates suggesting hyperscalers will raise another $400 billion in bonds in 2027.
• A key warning is that escalating costs, including spiking interest rates and NVIDIA’s 15% price hike on GPUs due to DRAM costs, are making the AI infrastructure buildout increasingly untenable, creating a “KobayAIshi Maru” scenario for companies deeply invested.