Paul Graham: Chase Power, Not Just Revenue, to Build a Breakout Startup

Paul Graham argues that the sharpest question a founder can ask isn’t how to make more money—that yields incremental gains—but how to make the company more powerful, which can multiply its value by orders of magnitude. He catalogs the levers he probes during YC office hours: owning the customer relationship rather than being a replaceable component, positioning yourself so money physically flows through you, and building platform dynamics (an app store, extensibility, or open APIs) so that other companies’ efforts compound your own value. Network effects are his recurring obsession; he treats it as a puzzle to engineer them into products that seem to have none, sometimes by generalizing a narrow service into a two-sided marketplace. A payments-for-AI-agents startup, for instance, becomes far more valuable the moment agents can pay each other.

Several of his heuristics turn on watching how a business actually behaves in the wild. ‘Tails that wag the dog’—like PayPal, which began as a security demo before eBay sellers repurposed it into a payments company—signal latent demand so strong that users will misuse a product to get what they want; founders should listen rather than resist. Helping customers make money accelerates both adoption and willingness to pay. And selling to early-stage startups, counterintuitively, compounds revenue at startup growth rates once usage-based pricing locks in low-churn infrastructure, which is why Stripe races to sign companies at their inception.

The essay’s broader thesis is that patience and generosity are structural advantages precisely because most competitors and corporate counterparts think only in quarters. Long-term tradeoffs get systematically underpriced, and giving away value—open-sourcing software to become a trusted standard, or heeding Tim O’Reilly’s dictum to create more value than you capture—tends to unlock 10x or 100x returns that penny-pinching optimization never will. For technical founders, the takeaways are concrete: default to open APIs (especially as AI agents replace human users and their needs become unpredictable), design for extensibility, and treat platform ownership as the real prize.