Anh-Tho Chuong, cofounder of Lago, highlights a significant challenge in AI-driven businesses: pricing models that traditionally worked are now failing. Many AI startups are experiencing growth that can lead to financial instability due to high costs associated with usage-based billing from large language model (LLM) providers. Unlike traditional SaaS, where costs remain stable as user numbers grow, AI companies face escalating expenses as usage increases. This has resulted in collapsing gross margins, as seen with Replit, which saw its margins plummet from 36% to negative 14% in just a few months. Chuong outlines five pricing models suited for the AI landscape. Usage-based pricing, where customers pay for what they consume, reflects how many infrastructure providers operate today. Seat-based subscriptions still have their place, especially for products that enhance collaboration, but AI's efficiency might reduce the number of necessary seats. The subscription model with overages allows companies to set a base price while charging extra for additional usage, ensuring that critical services remain uninterrupted. Understanding these models is essential for founders and teams, as pricing impacts not just revenue but the overall business strategy. With competition intensifying and profit margins shrinking, a company-wide focus on pricing and monetization is necessary. Founders must adapt quickly to turn the high costs of AI into a competitive advantage rather than a liability.