Premium: What If...We're In An AI Bubble? (Part 2)

Part 2 of an ongoing series examining the possibility of an AI bubble, exploring cascading consequences if data center construction slows, including NVIDIA GPU oversupply, hyperscaler revenue shortfalls, and $178.5B in data center debt going unpaid. The piece argues that apparent AI compute demand is largely an illusion created by Anthropic and OpenAI absorbing available capacity, while broader enterprise demand has plateaued. It critiques the circular dependency between AI labs, venture capital, and AI startups — where everyone assumes someone else will solve the profitability problem. The installment previews scenarios around VC funding drying up, AI startups collapsing, and inference remaining unprofitable.