For the entire history of software, how much one person could use was capped by a single number. Hours at the keyboard. One screen, one task, maybe eight focused hours a day if you're really grinding. Claude Code Projects removes that cap, and the economics underneath are wild. Here's the mechanism. A project is one standing conversation where Claude splits your work into threads and runs each as a full cloud session in parallel. You send a thought, close the laptop, and five sessions keep working. Boris Cherny, who created Claude Code, says he stopped managing sessions entirely and just sends ideas as they come. The work gets divided without him. That reads like a convenience feature. Now look at what Anthropic's own docs warn you about. Projects "use plan limits faster" because every thread is a complete Claude Code session. Ten threads running overnight means one subscriber pulling ten users' worth of compute while asleep. Demand per user used to be bounded by attention. Now it's bounded by ambition, and ambition doesn't sleep. Which explains the waitlist. Nobody capacity-gates a chat feature. You capacity-gate a feature where a single $200/mo account can suddenly consume like a small team. Every flat-rate software subscription ever sold was priced against the same assumption, that a human can only sit at a screen so many hours a day. That assumption lasted fifty years. It didn't survive this one.